Large-load high-speed long-voyage anti-interference unmanned aerial vehicle system

Abstract

The invention belongs to the technical field of unmanned aerial vehicles, and particularly relates to a large-load high-speed long-voyage anti-interference unmanned aerial vehicle system which comprises a high-strength light-weight vehicle body module, a four-rotor high-power power unit, a multi-mode redundant navigation module, a self-adaptive anti-interference communication link module, an embedded AI task processing unit and a high-energy-density power supply unit. The high-strength light-weight aircraft body module has the dual advantages that the aircraft body is integrally formed through carbon fibers, the Y-shaped main beam is arranged, and after taking off, the aircraft body is automatically locked to form a rigid whole, so that the mass of the whole aircraft is greatly reduced, and the energy utilization efficiency is improved; in addition, under the high dynamic working conditions such as large-load hovering, high-speed forward flying or strong wind disturbance, flexible deformation of the vehicle arms is effectively restrained, the lifting force eccentricity, attitude oscillation and even out-of-control risks caused by structure looseness are avoided, and therefore the structural robustness, aerodynamic efficiency and flight control precision of the whole vehicle are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) technology, specifically a high-payload, high-speed, long-range, anti-interference UAV system. Background Technology

[0002] Unmanned aerial vehicles (UAVs) are aircraft that do not require human pilots to operate on board and rely on remote control or autonomous flight programs to complete specific tasks. With the development of artificial intelligence, materials science and communication technology, modern UAVs have been widely used in military reconnaissance, logistics transportation, emergency rescue, agricultural plant protection and geographic surveying. Especially in complex environments or high-risk scenarios, high-performance UAVs are gradually becoming critical mission execution platforms due to their flexibility, efficiency and scalability.

[0003] Driven by both modern aerospace technology and the demands of multi-field applications, unmanned aerial vehicles (UAVs) have evolved from simple reconnaissance and mapping tools into multi-functional platforms with "large payload, high speed, long range, and anti-interference capabilities." Currently, key scenarios such as emergency rescue, border patrol, maritime rights protection, and long-distance logistics place stringent demands on the comprehensive performance of UAVs. In emergency rescue, the rapid delivery of relief supplies and medical equipment to remote disaster sites requires UAVs to have large payload capacity and high-speed maneuverability. In border patrol and maritime rights protection scenarios, long-range endurance is the core of ensuring mission coverage, while complex electromagnetic environments (such as radar interference and communication shielding) place rigid demands on the anti-interference capabilities of UAVs. Similarly, scenarios such as long-distance logistics and cross-regional material delivery require UAVs to balance large payload and long range while resisting adverse factors such as electromagnetic interference and airflow disturbances in complex environments to ensure stable mission execution.

[0004] Despite significant progress in current drone technology, it still faces multiple challenges in terms of overall performance, including high payload, high speed, long range, and strong anti-jamming capabilities. These challenges are specifically manifested in the following ways: First, it is difficult to balance structural strength and lightweight design. Traditional airframes often use modular assembly or ordinary composite materials, resulting in insufficient rigidity. Under heavy loads or high-speed maneuvers, they are prone to deformation or even structural failure. Existing folding arm designs lack active locking mechanisms, failing to form a rigid whole during flight, affecting aerodynamic efficiency and control response. Second, navigation systems lack reliability in complex environments. Most UAVs rely on GNSS signals, which are prone to loss of lock in urban canyons, indoor environments, or environments with electromagnetic interference. Single inertial or visual navigation suffers from large cumulative errors and altitude drift, lacking multi-source tight coupling and automatic switching mechanisms, leading to a sharp drop in positioning accuracy or even loss of control. Third, communication links have weak anti-interference capabilities and rigid resource scheduling. Conventional UAVs typically use fixed frequency band communication, which is prone to interruption in scenarios with strong interference or obstruction. At the same time, control commands and video streams share the same channel, and bandwidth and redundancy strategies are not dynamically allocated according to task priority, causing delays in critical commands or video stuttering, seriously affecting the continuity and safety of mission execution. Therefore, this paper proposes a high-payload, high-speed, long-range, anti-interference UAV system to address the above problems. Summary of the Invention

[0005] To overcome the shortcomings of existing technologies and solve at least one of the technical problems mentioned in the background art, this invention proposes a high-payload, high-speed, long-range, anti-interference unmanned aerial vehicle system.

[0006] The technical solution adopted by the present invention to solve its technical problem is: a high-load, high-speed, long-range anti-interference unmanned aerial vehicle system, comprising a high-strength lightweight airframe module, a quadcopter high-power power unit, a multimodal redundant navigation module, an adaptive anti-interference communication link module, an embedded AI task processing unit, and a high-energy-density power supply unit. The high-strength lightweight airframe module has a built-in Y-shaped main beam, and two sets of arms are rotatably connected to the fuselage via hinges with electromagnetic clutches. The electromagnetic clutches are electrically connected to the flight control unit, which controls the arms to form a rigid coaxial structure after takeoff. The quadcopter high-power power unit includes multiple motors and propellers. The multiple motors are connected to the flight control unit through an electronic speed controller that supports bidirectional communication, which is used to receive speed commands and transmit motor status back. The multimodal redundant navigation module is connected to the flight control unit via communication; the flight control unit runs the VIO-GNSS tightly coupled algorithm, automatically switches to the navigation mode dominated by visual inertial odometry when the GNSS signal fails, and uses laser ranging data to correct the cumulative error in the altitude direction. The adaptive anti-interference communication link module is connected to the communication management unit. The communication management unit detects the status of the optical fiber communication link in real time, evaluates the quality of each wireless communication link, and dynamically selects the primary wireless communication channel based on task priority. The embedded AI task processing unit receives data from the environmental perception sensor, runs the target detection and tracking algorithm, generates dynamic waypoints, and sends them to the flight control unit via UDP protocol. The high-energy-density power unit is equipped with a temperature sensor connected to the battery management system (BMS), which communicates with the flight control unit.

[0007] Preferably, the multimodal redundant navigation module includes an RTK-GNSS receiver, a high-precision six-axis IMU, a global shutter binocular camera, a single-point laser rangefinder, a digital barometer, and a three-axis geomagnetic sensor.

[0008] Preferably, the communication management unit detects the online / offline status of the fiber optic communication link in real time, and monitors the signal-to-noise ratio, packet loss rate, round-trip time, and bandwidth utilization of each wireless communication link in real time. When the fiber optic communication link is detected to be online, the fiber optic communication link is selected; when the fiber optic communication link is detected to be offline, the control command is routed to the wireless communication link with the lowest latency. The video stream adopts a segmented parallel redundant transmission mechanism, and key telemetry data is backed up and sent across links to ensure flight safety and mission continuity under conditions of strong electromagnetic interference, multipath fading, or partial link interruption.

[0009] Preferably, the flight control unit establishes bidirectional communication with each functional module via a high-speed digital bus: The battery management system is connected to the multimodal redundant navigation module and the high energy density power supply unit via a CAN bus; It is directly connected to the digital barometer and triaxial magnetometer in the multimodal redundant navigation module via SPI bus to obtain high-frequency inertial data; it is also connected to the multimodal redundant navigation module to obtain high-frequency inertial data from the digital barometer and triaxial magnetometer. It receives waypoint instructions encapsulated in UDP via Ethernet or high-speed UART from the embedded AI task processing unit; Exchange link status and priority strategies with the adaptive anti-interference communication link module via a serial interface or a dedicated communication management channel; The flight control unit also outputs control signals to the electromagnetic clutch drive circuit in the high-strength lightweight airframe module to achieve dynamic control of arm stiffness.

[0010] Preferably, the Y-shaped main beam integrates a deployable aerodynamic fairing, which is made of shape memory alloy and deployed via the flight control unit. The fairing automatically retracts during takeoff, landing, or low-speed maneuvering to avoid interfering with the propeller downwash airflow.

[0011] Preferably, the embedded AI task processing unit supports multi-task parallel reasoning, including but not limited to target recognition, semantic map construction, dynamic obstacle prediction, and emergency return path replanning; When the communication link quality is detected to exceed the threshold, it automatically switches to local closed-loop control mode, generates an obstacle avoidance trajectory based on the latest environmental perception data, and performs autonomous return.

[0012] Preferably, the high-energy-density power unit is equipped with a dual-battery hot-swappable interface and supports dynamic power balancing during flight; The battery management system monitors the voltage, internal resistance, and temperature rise gradient of each cell in real time. When any cell malfunctions, the flight control unit immediately isolates the faulty module and switches to the backup power circuit, while simultaneously sending a battery health status alarm to the ground station.

[0013] Preferably, the multimodal redundant navigation module further integrates millimeter-wave radar for detecting the distance and relative speed of obstacles ahead in dense fog, sandstorms, or low-light conditions at night; Radar point cloud data is fused with binocular visual features and input into a robust state estimator in the flight control unit to enhance local obstacle avoidance and navigation capabilities in VIO-GNSS failure scenarios.

[0014] Preferably, the communication management unit dynamically selects network access points to ensure the continuous availability of public network links during long-distance cross-regional missions.

[0015] Preferably, the flight control unit is equipped with a fail-safe state machine, which is a hardware and software collaborative fault monitoring and hierarchical response control mechanism integrated within the flight control unit. By collecting the status data of each functional module in real time, it constructs the state switching logic of normal operation, early warning, fault and emergency. When it is detected that more than two main sensors in the multimodal redundant navigation module have failed or the communication link has been completely interrupted for more than a preset time limit, the emergency response mechanism is automatically triggered.

[0016] The beneficial effects of this invention are: This invention provides a high-payload, high-speed, long-range, anti-interference unmanned aerial vehicle (UAV) system. It employs a high-strength, lightweight airframe module with a unibody carbon fiber fuselage and an integrated Y-shaped main beam. Innovatively, it introduces an electromagnetic clutch hinge mechanism electrically controlled by the flight control unit at the connection between the arms and fuselage. This achieves the dual advantages of easy storage and transportation before takeoff and automatic locking to form a rigid whole after takeoff. This design not only significantly reduces the overall weight and improves energy efficiency but also effectively suppresses flexible deformation of the arms under high-dynamic conditions such as heavy-load hovering, high-speed forward flight, or strong wind disturbances. It avoids the risks of lift eccentricity, attitude oscillation, or even loss of control caused by structural loosening, thereby significantly enhancing the overall structural robustness, aerodynamic efficiency, and flight control precision.

[0017] This invention provides a high-payload, high-speed, long-range, anti-jamming unmanned aerial vehicle (UAV) system. By deploying a multimodal redundant navigation module, integrating GNSS, visual inertial odometry (VIO), and laser ranging sensors, and running a tightly coupled VIO-GNSS fusion algorithm in the flight control unit, a multi-layered, adaptive navigation architecture is constructed. When the UAV enters urban canyons, indoor environments, tunnels, or encounters GNSS spoofing or suppression interference, the system can automatically switch to a passive navigation mode dominated by VIO within milliseconds. Simultaneously, it utilizes high-precision laser ranging data to correct vertical attitude integral drift in real time, effectively suppressing altitude divergence under prolonged GNSS-free conditions. This mechanism significantly improves positioning continuity and accuracy, ensuring reliable navigation capabilities at the centimeter to decimeter level even in signal-denied areas during long-range missions, fundamentally solving the risk of traditional UAVs losing contact or control in complex electromagnetic or obstructed environments.

[0018] This invention provides a high-payload, high-speed, long-range, anti-jamming unmanned aerial vehicle (UAV) system. By integrating an adaptive anti-jamming communication link module and an intelligent communication management unit, the system can monitor the status of fiber optic communication links and the signal-to-noise ratio, packet loss rate, and latency of multiple wireless communication channels (such as 2.4GHz, 5.8GHz, L-band, or 4G / 5G auxiliary links) in real time. Based on task semantic priority, it implements dynamic resource scheduling: high-security flight control commands are routed to the primary link with the lowest latency and highest reliability, while high-definition video streams employ segmented encoding and multi-path redundant transmission mechanisms, allowing for image reconstruction even when some links are interfered with. Furthermore, the system supports anti-jamming strategies such as frequency hopping, spread spectrum, or encrypted retransmission, significantly improving link resilience. This design not only ensures the real-time and deterministic nature of remote control but also guarantees the continuity and availability of video transmission during reconnaissance and inspection missions, significantly enhancing the UAV's mission execution and survivability in highly contested or complex terrain environments. Attached Figure Description

[0019] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention.

[0020] In the attached diagram: Figure 1 This is a system flowchart of the present invention; Figure 2 This is a schematic diagram of the multimodal redundant navigation module in this invention; Figure 3 This is a schematic diagram of the adaptive anti-interference communication link module in this invention; Figure 4 This is a flowchart of the flight control unit in this invention. Detailed Implementation

[0021] The technical solutions of 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.

[0022] Specific implementation examples are given below.

[0023] Please see Figures 1-4 This invention provides a high-payload, high-speed, long-range, anti-interference unmanned aerial vehicle system, including a high-strength lightweight airframe module, a quadcopter high-power power unit, a multimodal redundant navigation module, an adaptive anti-interference communication link module, an embedded AI task processing unit, and a high-energy-density power supply unit. All modules are uniformly scheduled and coordinated in a closed loop through a flight control unit. After system startup, the flight control unit first completes a Power-On Self-Test (POST) of each subsystem to confirm sensor status, battery health, communication link availability, and normal arm locking mechanism function. During takeoff, the arms are in a foldable state to reduce ground space occupation; after takeoff, the flight control unit immediately outputs a high-level signal to the electromagnetic clutch drive circuit, energizing and engaging it to rigidly lock the two sets of arms to both sides of the Y-shaped main beam, forming a high-rigidity flight configuration. At the same time, the fairing retracts to avoid interfering with the propeller downwash airflow. During the cruise phase, the flight control unit uses dynamic waypoints generated by the embedded AI task processing unit and multimodal navigation information to perform high-precision trajectory tracking, and ensures reliable transmission of commands and video streams through adaptive communication management strategies. When encountering abnormal conditions such as GNSS denial, strong electromagnetic interference, or communication interruption, the system automatically activates redundancy mechanisms, switching to vision-led navigation, local closed-loop control, or a three-level emergency landing procedure to ensure flight safety.

[0024] In some embodiments, such as Figures 1-2 As shown, the high-strength, lightweight body module: The high-strength, lightweight body module is manufactured using T700 grade carbon fiber prepreg through a one-piece molding process in an autoclave, with the overall weight controlled to within 1.8 kg. The fuselage integrates a Y-shaped main beam, which is CNC machined from 7075-T6 aviation aluminum. As the main load-bearing structure, its three-way intersection point is located directly above the center of gravity to ensure torque balance when the load is suspended. The two sets of arms are connected to the main beam via precision hinges with normally open electromagnetic clutches (model: MCE-24V-5A, holding torque ≥15 N·m). The electromagnetic clutch coils are controlled by the GPIO pins of the flight control unit via a MOSFET driver circuit (IRF540N). When energized, they generate magnetic force to press the friction discs together, achieving mechanical locking. The locked state is fed back to the flight control unit via a Hall switch, forming a closed-loop confirmation.

[0025] The mission payload is mounted via a quick-release center sling. The position of the sling is optimized through finite element simulation to ensure that its gravity line is coaxial with the resultant lift line of the quadrotor within ±2 mm, thus avoiding yaw moment.

[0026] Furthermore, a deployable aerodynamic fairing is integrated into the inner cavity of the leading edge of the Y-shaped main beam, driven by nickel-titanium-based shape memory alloy (SMA) wires. The two ends of the SMA wires are fixed to the fairing's pivot and the main beam support. After being heated to the austenitic phase transformation temperature (approximately 85°C), the fairing contracts, pulling it to rotate 90° around the pivot and unfold, covering the motor and arm areas. The fairing surface is coated with a low-friction polytetrafluoroethylene coating, which reduces the cruise drag coefficient by up to 18% after unfolding. The flight control unit triggers the unfolding command based on airspeed sensor readings (>15 m / s) and monitors the SMA temperature via an NTC thermistor to prevent overheating failure.

[0027] In some embodiments, such as Figures 2-3 As shown, the configuration and control of the high-power quadcopter power unit: The power unit uses four T-Motor U15 KV80 brushless motors, with a maximum continuous power of 3.2 kW per unit, paired with 32×10-inch carbon fiber large-diameter propellers, providing a total thrust margin of ≥2.5 times the hovering requirement. Each motor is connected to the BLHeli_32 firmware electronic speed controller (ESC) via the DShot1200 protocol (digital PWM, baud rate 1.2 Mbps) to achieve microsecond-level response.

[0028] The flight control unit sends the target rotation speed command to the ESC every 2 ms. and receive the real-time rotation speed transmitted back. Bus voltage and temperature The speed error is used to build the underlying PID controller: in =1.2, =0.05, =0.15, the parameter was tuned using the Ziegler-Nichols method and verified in the wind tunnel.

[0029] To suppress high-frequency vibration, the flight controller superimposes a notch filter onto the motor command channel after attitude calculation, with the center frequency set to the first-order bending mode of the airframe (measured ≈ 42 Hz) and Q value = 10.

[0030] Data fusion algorithm for multimodal redundant navigation modules: The navigation module includes the following sensors: RTK-GNSS receiver (u-blox F9P), outputs WGS84 coordinates, update rate 10 Hz, horizontal accuracy ±1 cm +1 ppm; Six-axis IMU (BMI088), accelerometer range ±24 g, gyroscope ±2000 ° / s, sampling rate 1000 Hz; Global shutter stereo camera (OV9282 ×2), baseline distance 80 mm, synchronization frame rate 30 Hz; Single-point laser rangefinder (TFmini Plus), range 0.1–12 m, accuracy ±6 cm; Digital barometer (MS5611), height resolution 10 cm; Triaxial magnetometer (HMC5983); Millimeter-wave radar (TI IWR6843ISK), 77 GHz, detection range 0.2–20 m, velocity accuracy ±0.1 m / s.

[0031] All sensors are synchronized via hardware triggering or PTP (Precision Time Protocol), with a maximum deviation of <100 μs.

[0032] The flight controller operates a VIO-GNSS tightly coupled filter, and the state vector is defined as follows: in , , For position, attitude (quaternion), and velocity; , Zero bias for IMU; This is the scale factor (used for monocular initialization; fixed at 1 in binocular mode).

[0033] The filtering framework adopts an improved version of MSCKF (Multi-State Constraint Kalman Filter), and the feature point constraint equation is as follows: in Let i be the camera pose in the i-th frame. For spatial points, (.) represents the pinhole projection model.

[0034] The GNSS observation model is as follows: in The antenna's position relative to the aircraft body (already calibrated). Rotate ENU to the world coordinate system.

[0035] Laser ranging provides height correction: in Let Z be the body coordinate. For laser mounting offset. This observation is used to suppress VIO vertical drift, with covariance set to... .

[0036] If the GNSS signal is lost for more than 3 seconds, the system automatically degrades to pure VIO mode and enables altitude estimation by fusion of barometer and laser signals. After clustering the millimeter-wave radar point cloud using DBSCAN, the distance to the nearest obstacle is extracted. With relative velocity The data is input into the local obstacle avoidance cost map for emergency deceleration decisions.

[0037] In some embodiments, such as Figures 1-3 As shown, the communication management unit is deployed on a separate ARM Cortex-M7 microcontroller, such as the STM32H743, with a clock frequency of 480 MHz, equipped with 1 megabyte of static random access memory, and running the FreeRTOS real-time operating system. This unit exchanges data bidirectionally with the flight control unit via a high-speed serial interface (such as UART, baud rate 921600 bits per second) or SPI bus. The communication management unit continuously monitors the fiber optic communication link status every 100 milliseconds and collects key performance indicators for three types of wireless communication links.

[0038] For the fiber optic communication module, the system reads its received optical power, transmitted optical power, bit error rate, and link connectivity status through the module's industrial-grade communication interface. The typical range of received optical power is -30 to 0 dBm, the typical range of transmitted optical power is 0 to 10 dBm, the bit error rate can be as low as 10^-12, and the link connectivity status is fed back in real time with high and low level signals. After the aforementioned parameters are parsed by the fiber optic communication module's underlying driver, they are transmitted to the communication management unit through standard data frames. For a 2.4 GHz frequency-hopping remote control link, the system reads its Received Signal Strength Indication (RSSI), Packet Error Rate (PER), and Round-Trip Time (RTT). RSSI typically ranges from -100 to -30 dB / mW, PER values ​​are between 0 and 1, and RTT is generally between 5 and 100 milliseconds. These data are directly provided by the internal registers of the frequency-hopping transceiver chip (such as the SX1280).

[0039] For a 5.8 GHz COFDM image transmission link, the system acquires the signal-to-noise ratio (SNR), bit error rate (BER), and Doppler shift. Typical SNR values ​​range from 0 to 30 dB, BER from 10⁻⁶ to 1, and Doppler shift can reach ±500 Hz. These parameters are returned by a COFDM baseband processing chip (such as the Amimon AMN2120) through a dedicated application programming interface (API).

[0040] For 4G / 5G cellular communication modules, the system queries the Reference Signal Received Power (RSRP), Signal-to-Interference-Noise Ratio (SINR), and end-to-end delay via AT commands or the QMI protocol. The RSRP ranges from -140 to -44 dBmW, the SINR is between -3 and 30 dB, and the delay is typically between 30 and 500 milliseconds. Based on the above indicators, the communication management unit monitors the online / offline status of the fiber optic communication links in real time and constructs a multi-dimensional weighted scoring function to quantify the current service quality of each wireless communication link. When the fiber optic communication link is detected as online, the communication management unit selects the fiber optic communication link; when the fiber optic communication link is detected as offline, the communication management unit selects the wireless communication link. The multi-dimensional weighted scoring formula is as follows: Where k represents the link number (1 for 2.4 GHz, 2 for 5.8 GHz, and 3 for 4G / 5G); ε is a small constant to prevent division by zero, set to 1 millisecond; The R normalization interval is set to a minimum of 5 dB and a maximum of 25 dB; default weights. It can also be remotely configured via ground station to adapt to different mission requirements.

[0041] Control commands (including critical information such as flight control telemetry data and emergency commands) are encapsulated into UDP datagrams of no more than 256 bytes and forcibly routed to the physical link with the highest score, i.e., selecting S... k The link corresponding to the largest k value. The system strictly guarantees that the end-to-end latency of this type of data is less than 20 milliseconds; if the latency of the optimal link exceeds the limit, it will automatically degrade to use the suboptimal link and send a link quality alarm to the ground station.

[0042] The video stream employs a fragmented redundancy transmission mechanism: the original H.265 encoded stream is divided into multiple logical segments according to a Group of Pictures (GOP) structure, typically each segment containing approximately 50 to 100 kilobytes of data. Two identical copies of each segment are generated and transmitted in parallel via two different physical links—for example, one via a 5.8 GHz video transmission and the other via a 4G network. The ground receiver buffers all arriving data segments and uses Reed-Solomon forward error correction codes (using RS(255,223) parameters) to recover lost content. If neither copy of a segment is received, temporal interpolation or a spatial repair strategy based on adjacent frames is employed to maintain image continuity.

[0043] The 4G / 5G module (such as the Quectel RM500Q-GL) has two physical SIM card slots and one eSIM chip, supporting the GSMA SGP.02 standard. The communication management unit downloads the operator configuration file from a remote SM-DP+ server via the Local Configuration Assistant (LPA) service. In multi-operator coverage areas, the system makes a comprehensive decision on the optimal access point based on signal quality and cost, with the following handover strategy: in This represents the unit traffic cost (in yuan per megabyte) for the i-th operator in the current geographic region. The cost sensitivity factor is set to 0.1 by default. The handover process uses the QUIC multipath extension protocol to achieve seamless session migration, with handover interruption time controlled within 300 milliseconds.

[0044] In some embodiments, such as Figures 1-4 As shown, the embedded AI task processing unit: The embedded AI task processing unit is built on the NVIDIA Jetson AGX Orin computing module, possessing 32 TOPS of INT8 integer operation capability, and runs the Ubuntu 22.04 operating system and the ROS 2 Humble middleware framework. After system startup, it loads the YOLOv8n (Nano version) object detection model optimized with TensorRT 8.6 and completed with INT8 quantization. The input image resolution is 640 x 640 pixels, and the inference time per frame is approximately 22 milliseconds, corresponding to a processing speed of 45 frames per second.

[0045] The detection results are output in structured data format, with each target including a category identifier, bounding box coordinates (x-center, y-center, width, and height), and a confidence score. The system only activates the DeepSORT multi-target tracker for targets with a confidence score higher than 0.7. This tracker maintains a state vector for each target, including position, aspect ratio, height, and its first derivative, and uses Kalman filtering for motion prediction. It also utilizes a 128-dimensional appearance feature embedding vector for cross-frame data association, effectively suppressing ID switching caused by occlusion or brief loss of data.

[0046] Dynamic waypoint generation employs a feedforward predictive control strategy: based on the target's current position, its velocity vector displacement over a preset look-ahead time is superimposed, and an overshoot coefficient is introduced to compensate for maneuverability. The specific formula is: in, Set it to 1.2. The time is 2.0 seconds. The waypoints generated in this way can effectively guide drones to intercept moving targets in advance; When the communication management unit reports link quality degradation (round-trip latency exceeding 500 milliseconds and packet error rate higher than 30%) for 5 consecutive seconds, the AI ​​unit automatically switches to local closed-loop autonomous mode. At this time, the system calls the improved RRT* (Rapid Exploration Random Treestar) algorithm to replan the return path online. The path cost function comprehensively considers the distance to the home point and the proximity to obstacles, the latter being modeled using an exponential decay function, with a safe distance threshold set at 1.5 meters.

[0047] The generated waypoints are broadcast to the flight control unit via UDP protocol, with the target address being 192.168.1.10 and port number 5005. The message body uses standard JSON format and includes a unique waypoint identifier, three-dimensional coordinates, yaw angle, and a nanosecond-level timestamp. Upon receiving the message, the flight control unit immediately adds it to the task queue and sends an acknowledgment response back via port 5006.

[0048] In some embodiments, such as Figures 1-4 As shown, the safety management mechanism of the high-energy-density power supply unit is as follows: The power system consists of two sets of parallel solid-state lithium battery modules. Each set uses a 16-series-2-parallel LiFePO4 cell architecture, paired with a lithium lanthanum zirconium oxide (LLZO) ceramic solid electrolyte. The nominal voltage is 52.8 volts, the total capacity is 32 amp-hours, the total system energy is 1.69 kilowatt-hours, and the mass energy density is approximately 320 watt-hours per kilogram. The battery management system is based on the STM32G474 microcontroller and integrates a 12-bit analog-to-digital converter. It synchronously acquires the voltage of each cell string (accuracy ±5 mV), the temperature points of eight thermistors (accuracy ±1 degree Celsius), and the total output current (measured by an ACS770 Hall sensor with a range of ±200 amperes and an accuracy of ±1%) at a 100-millisecond cycle. The power balancing adopts a passive method: when the voltage difference between any two strings of cells exceeds 50 millivolts, the system activates the corresponding discharge resistor (1 ohm resistance, 5 watts power), and the balancing current is limited to within 2 amps until the voltage difference is reduced to below 20 millivolts. The fault isolation logic uses software to determine faults in real time: if any cell voltage is below 2.8 volts (undervoltage) or above 3.65 volts (overvoltage), or the temperature exceeds 65 degrees Celsius, or the temperature rise rate is greater than 5 degrees Celsius per second (signs of thermal runaway), the corresponding MOSFET for that cell group is immediately shut down, and a fault flag is set. If the fault occurs in the main battery pack, the system automatically switches to the backup battery pack for power supply and simultaneously sends a fault alarm to the flight controller via the CAN bus. The flight control unit listens to the status broadcast frames (identifier 0x201) of the battery management system via a 500 kilobits per second CAN bus. The frames contain the total voltage (in 0.1 volts), the remaining charge percentage, the fault mask (each bit corresponds to a fault type), and the highest temperature value.

[0049] When the battery temperature exceeds 60 degrees Celsius, the flight controller implements three levels of thermal protection measures: first, it limits the motor output power to 80% of the rated value; second, it starts the 24-volt turbofan at the bottom of the fuselage, whose pulse width modulation duty cycle increases linearly with temperature (10% for every 1 degree Celsius increase, up to 100%); and finally, it sends a "battery overheating" alarm message to the ground station.

[0050] In some embodiments, such as Figures 1-4 As shown, the fault-safe state machine design of the flight control unit is as follows: The flight control software is based on PX4 Autopilot 1.14 and is deeply customized. A new three-level fail-safe state machine has been added, with an independent watchdog thread monitoring the trigger conditions at 10-millisecond intervals and executing the corresponding actions.

[0051] The first level is the hovering state, which is triggered by the following conditions: the global navigation satellite system and visual inertial odometry fail simultaneously, or the communication interruption time is less than 10 seconds. At this time, the flight control switches to the altitude control mode to maintain the current position and flight altitude. If this state lasts for more than 15 seconds, it will automatically enter the second level. The second stage is a controlled descent, which requires the following conditions to be met simultaneously: the first stage has lasted for more than 15 seconds, and the binocular cameras have successfully extracted no fewer than 50 stable visual feature points. The flight controller descends at a constant rate of 0.5 meters per second, and the horizontal position is estimated by fusion of the optical flow sensor and the inertial measurement unit. When the flight altitude is below 2 meters, the landing procedure is executed; if communication is restored during this period, the system returns to normal mission mode. The third level is the emergency landing state, triggered when: neither the laser rangefinder nor the barometer can provide valid altitude information, or the power output drops by more than 50% within one second. In this case, the available altitude source for flight control fusion (laser is preferred) executes an exponential decay descent strategy, and its altitude command is: Where the initial height is The time constant τ is set to 8 seconds. When the descent speed is below 0.1 meters per second for 1 consecutive second, it is determined that the ground has been reached, and all motors are immediately shut down. The black box uses a Fujitsu MB85RS2MT ferroelectric memory chip, connected via an SPI interface, to cyclically record critical system data for the past 30 seconds at a frequency of 100 Hz. This data includes raw inertial sensor values, motor control commands, navigation status estimates, communication link quality, and battery health parameters. The memory is designed to retain data for up to ten years without power loss, and ground maintenance personnel can download the logs via USB or wirelessly for accident analysis.

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

Claims

1. A high-payload, high-speed, long-range, anti-jamming unmanned aerial vehicle (UAV) system, characterized in that: It includes a high-strength lightweight airframe module, a quadcopter high-power power unit, a multimodal redundant navigation module, an adaptive anti-interference communication link module, an embedded AI task processing unit, and a high-energy-density power supply unit; The high-strength lightweight airframe module has a built-in Y-shaped main beam, and two sets of arms are rotatably connected to the fuselage via hinges with electromagnetic clutches. The electromagnetic clutches are electrically connected to the flight control unit, which controls the arms to form a rigid coaxial structure after takeoff. The quadcopter high-power power unit includes multiple motors and propellers. The multiple motors are connected to the flight control unit through an electronic speed controller that supports bidirectional communication, which is used to receive speed commands and transmit motor status back. The multimodal redundant navigation module is connected to the flight control unit via communication; the flight control unit runs the VIO-GNSS tightly coupled algorithm, automatically switches to the navigation mode dominated by visual inertial odometry when the GNSS signal fails, and uses laser ranging data to correct the cumulative error in the altitude direction. The adaptive anti-interference communication link module is connected to the communication management unit. The communication management unit detects the status of the optical fiber communication link in real time, evaluates the quality of each wireless communication link, and dynamically selects the primary wireless communication channel based on task priority. The embedded AI task processing unit receives data from the environmental perception sensor, runs the target detection and tracking algorithm, generates dynamic waypoints, and sends them to the flight control unit via UDP protocol. The high-energy-density power unit is equipped with a temperature sensor connected to the battery management system (BMS), which communicates with the flight control unit.

2. The high-payload, high-speed, long-range, anti-jamming unmanned aerial vehicle system as described in claim 1, characterized in that: The multimodal redundant navigation module includes an RTK-GNSS receiver, a high-precision six-axis IMU, a global shutter binocular camera, a single-point laser rangefinder, a digital barometer, and a three-axis geomagnetic sensor.

3. The high-payload, high-speed, long-range, anti-jamming unmanned aerial vehicle system as described in claim 1, characterized in that: The communication management unit detects the online / offline status of the fiber optic communication link in real time, and monitors the signal-to-noise ratio, packet loss rate, round-trip time, and bandwidth utilization of each wireless communication link in real time. When the fiber optic communication link is detected to be online, the fiber optic communication link is selected; when the fiber optic communication link is detected to be offline, the control command is routed to the wireless communication link with the lowest latency. The video stream adopts a segmented parallel redundant transmission mechanism, and key telemetry data is backed up and sent across links to ensure flight safety and mission continuity under conditions of strong electromagnetic interference, multipath fading, or partial link interruption.

4. The high-payload, high-speed, long-range, anti-jamming unmanned aerial vehicle system as described in claim 1, characterized in that: The flight control unit establishes bidirectional communication with each functional module via a high-speed digital bus: The battery management system is connected to the multimodal redundant navigation module and the high energy density power supply unit via a CAN bus; It is directly connected to the digital barometer and triaxial magnetometer in the multimodal redundant navigation module via SPI bus to obtain high-frequency inertial data; it is also connected to the multimodal redundant navigation module to obtain high-frequency inertial data from the digital barometer and triaxial magnetometer. It receives waypoint instructions encapsulated in UDP via Ethernet or high-speed UART from the embedded AI task processing unit; Exchange link status and priority strategies with the adaptive anti-interference communication link module via a serial interface or a dedicated communication management channel; The flight control unit also outputs control signals to the electromagnetic clutch drive circuit in the high-strength lightweight airframe module to achieve dynamic control of arm stiffness.

5. The high-payload, high-speed, long-range, anti-jamming unmanned aerial vehicle system as described in claim 1, characterized in that: The Y-shaped main beam integrates a deployable aerodynamic fairing, which is made of shape memory alloy and deployed via the flight control unit. It automatically retracts during takeoff, landing, or low-speed maneuvering to avoid interfering with the propeller downwash airflow.

6. The high-payload, high-speed, long-range, anti-jamming unmanned aerial vehicle system as described in claim 1, characterized in that: The embedded AI task processing unit supports multi-task parallel reasoning, including but not limited to target recognition, semantic map construction, dynamic obstacle prediction, and emergency return path replanning. When the communication link quality is detected to exceed the threshold, it automatically switches to local closed-loop control mode, generates an obstacle avoidance trajectory based on the latest environmental perception data, and performs autonomous return.

7. The high-payload, high-speed, long-range, anti-jamming unmanned aerial vehicle system as described in claim 1, characterized in that: The high-energy-density power unit is equipped with dual-battery hot-swappable interfaces and supports dynamic power balancing during flight. The battery management system monitors the voltage, internal resistance, and temperature rise gradient of each cell in real time. When any cell malfunctions, the flight control unit immediately isolates the faulty module and switches to the backup power circuit, while simultaneously sending a battery health status alarm to the ground station.

8. The high-payload, high-speed, long-range, anti-jamming unmanned aerial vehicle system as described in claim 1, characterized in that: The multimodal redundant navigation module further integrates millimeter-wave radar, which is used to detect the distance and relative speed of obstacles ahead in dense fog, sandstorms, or low-light conditions at night. Radar point cloud data is fused with binocular visual features and input into a robust state estimator in the flight control unit to enhance local obstacle avoidance and navigation capabilities in VIO-GNSS failure scenarios.

9. A high-payload, high-speed, long-range, anti-jamming unmanned aerial vehicle system as described in claim 1, characterized in that: The communication management unit dynamically selects network access points to ensure the continuous availability of public network links during long-range cross-regional missions.

10. A high-payload, high-speed, long-range, anti-jamming unmanned aerial vehicle system as described in claim 1, characterized in that: The flight control unit is equipped with a fail-safe state machine, which is a hardware and software collaborative fault monitoring and hierarchical response control mechanism integrated within the flight control unit. By collecting the status data of each functional module in real time, it constructs the state switching logic of normal operation, early warning, fault and emergency. When it is detected that more than two main sensors in the multimodal redundant navigation module have failed or the communication link has been completely interrupted for more than a preset time limit, the emergency response mechanism is automatically triggered.

Images