Autonomous switching and multi-machine cooperative control system for cross-medium low-altitude unmanned aerial vehicles

Abstract

The invention discloses a cross-medium autonomous switching and multi-aircraft cooperative control system for low-altitude unmanned aerial vehicles, which belongs to the technical field of low-altitude aircrafts and consists of a multi-modal sensing unit, an autonomous switching unit and a cooperative control unit. The multi-modal sensing unit realizes heterogeneous data calibration through a space-time synchronization fusion algorithm; the autonomous switching unit generates a switching decision based on the six-dimensional environment feature vector; and the cooperative control unit integrates a dynamic contract network protocol and a medium adaptive path planning algorithm, and compresses task response delay to be less than or equal to 80ms. The technical bottlenecks of cross-medium sensing distortion, high switching decision false triggering rate and large multi-machine cooperation delay are solved, and the method can be well suitable for urban low-altitude pollution monitoring, underwater rescue and underground pipe network emergency operation.

Description

Technical Field

[0001] The present invention relates to the technical field of low-altitude aircraft, and in particular to an unmanned aerial vehicle control system integrating cross-media autonomous switching, multi-source environment perception and distributed collaborative decision-making functions. Background Art

[0002] As the application of unmanned aerial vehicles (UAVs) in urban 3D space monitoring, water rescue, and underground pipeline inspections deepens, the need for reliable switching and efficient multi-machine collaboration in complex media environments becomes increasingly prominent. Traditional cross-media UAV systems rely on a single threshold trigger mechanism (e.g., water depth > 1m) or centralized task scheduling. These systems face three major technical bottlenecks in complex operating conditions such as highly dynamic tides, turbulent disturbances, and confined underground spaces: (1) Insufficient perception reliability: The data failure rate of heterogeneous sensors (millimeter wave radar / sonar / LiDAR) is greater than 25% during medium switching (e.g., sonar failure in the air and radar failure underwater); (2) Switching decision rigidity: the fixed threshold method has a false trigger rate of >30% in tidal fluctuation scenarios; (3) Poor collaborative real-time performance: Centralized scheduling results in task response delays greater than 500ms at a scale of 20 nodes.

[0003] To date, research on cross-media unmanned systems still has significant limitations: (1) Patent CN114802751A “Water-air cross-medium unmanned vehicle and its control method” proposed a propulsion mode switching scheme based on water depth threshold. This scheme did not consider the effect of density gradient (▽ρ) on fluid dynamics. Its implementation showed that the propulsion efficiency decreased by 38.7% in tidal flow fields (density gradient ▽ρ ≥ 1.5 kg / m³); (2) Patent CN113064438B "Multi-UAV Collaborative Path Planning Method and System" adopts a centralized task allocation mechanism. Simulation shows that the probability of master node failure and crash reaches 43.6% at a scale of 20 nodes, and there is a lack of underground pipe network topology constraints; (3) The paper “Study on Multi-physics Field Coupling Modeling of Trans-media Aircraft” by Chen Tao et al. pointed out that magnetohydrodynamic propulsion is prone to cavitation effect during transient switching, but it did not solve the power interruption problem caused by the measured duct sealing response delay of more than 100ms in patent CN116136559A “A Sealing Device for Water-to-Air Trans-media Aircraft”.

[0004] Therefore, an innovative cross-media control system is urgently needed. It should address perception distortion through a multi-source spatiotemporal fusion algorithm (error suppression to <5%), overcome switching reliability bottlenecks based on a six-dimensional dynamic decision-making model (confidence ≥ 98%), and achieve efficient multi-machine collaboration through a distributed collaborative architecture (response latency ≤ 80ms). This system should meet the stringent requirements for cross-media adaptability, confined space maneuverability, and real-time emergency response in urban 3D space operations. Summary of the Invention

[0005] The main purpose of the present invention is to provide a cross-medium low-altitude unmanned aerial vehicle autonomous switching and multi-machine collaborative control system, which can effectively solve the problems mentioned in the background technology.

[0006] To achieve the above object, the technical solution adopted by the present invention is: A cross-medium low-altitude unmanned aerial vehicle autonomous switching and multi-machine collaborative control system includes a multimodal sensing unit, an autonomous switching unit, and a collaborative control unit, specifically comprising the following steps: S01. Build a multimodal perception unit, deploying a 77GHz millimeter-wave radar, a 200kHz multi-beam sonar, and a 905nm solid-state LiDAR in the main body of the aircraft. The sensors are fixed with titanium alloy brackets, and the sensor cables are connected to the main control board through the cable management trough inside the cabin. S02. Use spatiotemporal synchronous fusion algorithm to process heterogeneous data and establish sensor data mapping matrix: ; The delay compensation , ; S03: Based on the federated Kalman filter, environmental parameters are integrated to construct a six-dimensional feature vector (water depth, density gradient, turbidity, wind speed, methane concentration, and satellite navigation carrier-to-noise ratio). The switching decision is generated through the SVM classifier (kernel function γ = 0.5). When the medium probability P ≥ 0.98 and satisfies: When triggering mode conversion; S04. Execute propulsion system reconstruction: Switching from air to underwater: close the duct structure within 50ms (sealing pressure ≥ 12kPa), and start the magnetohydrodynamic thruster (thrust 5.5N±0.3N); Switching from underground to air: deploy the ducted fan (power ≥ 850W) and activate multi-mode satellite navigation / INS combined positioning (B1C+B2a dual-frequency); S05. Use the improved contract network protocol to achieve multi-machine collaboration, including dynamic link management of heterogeneous communication systems, and the task allocation response time t satisfies: ; Where d is the mission distance, v_max is the maximum speed of the UAV; S06, driving medium adaptive path planner: Air mode: RRT*-Smart algorithm integrated with turbulence cost model (path cost increases by 30% when wind speed > 8m / s); Underwater mode: The A*-NS algorithm is embedded in the Navier-Stokes equation solver, and the cost function is: ; Underground mode: The Topo-Dijkstra algorithm is subject to the pipe network topology constraints (minimum turning radius R ≥ 0.4m).

[0007] Preferably, the anti-interference hardware architecture design of the multimodal perception unit includes: The millimeter-wave radar is installed at the front of the aircraft, the sonar array is arranged in a waterproof cabin at the bottom (protection level IP68), and the LiDAR is placed on the rotating gimbal at the top; The magnetohydrodynamic propulsion system and the duct structure adopt a modular design and are connected through a quick-release interface. The duct surface is covered with a super-hydrophobic coating (contact angle > 150°). The main control board integrates a three-layer electromagnetic shield, the sensor cable uses shielded twisted pair cables, and the interface uses gold-plated shielded connectors; Each sensor is equipped with four silicone shock-absorbing pads (hardness Shore A50) and installed in a rectangular shape.

[0008] Preferably, the dynamic switching decision method of the autonomous switching unit includes: S31, the federated filter sets three sub-filters: Air sub-filter: processes millimeter-wave radar data (update frequency 100Hz) and air pressure data; Water medium sub-filter: Fusion of sonar data (accuracy ±0.05m@5m) and fluid pressure data; Subsurface subfilter: integrates LiDAR point cloud (scan rate 30Hz) and IMU data; S32. The main filter of the federated filter realizes cross-domain fusion through the improved covariance crossover algorithm, and outputs a state vector containing the medium type probability P_m∈[0,1] and the interface dynamic index D_i∈R⁺; The S33 and SVM classifiers use the RBF kernel function (γ = 0.5), and the training data set contains three types of medium boundary samples: tidal waters, turbulent air, and underground pipe networks.

[0009] Preferably, the multi-machine collaborative control method of the collaborative control unit includes: S41. Dynamic role allocation mechanism based on comprehensive scoring function: ; Where Q is the remaining power, D is the relative target distance, S res is the sensor resolution; S42, distributed fault-tolerant protocol execution process: Fault detection: If the heartbeat packet timeout is greater than 100ms, the node is considered to be faulty; Task redistribution: completed within 200ms through the Paxos consensus algorithm; Path replanning: RRT*-Smart algorithm convergence time < 500ms.

[0010] Preferably, the heterogeneous communication system includes: S51, the main link uses 5G private network (20MHz bandwidth), and the end-to-end delay is ≤10ms; S52, the backup link uses a combination of blue-green laser communication (wavelength 532nm, rate 2Mbps) and underwater acoustic communication (center frequency 50kHz); S53, communication protocol conversion process: Switching from air to underwater: interrupt the 5G link and activate the underwater acoustic communication module; Switch from underwater to air: turn off the sonar modulator and activate the laser transmitter; S54. The data security layer adopts SM4 national encryption algorithm for encryption (key update cycle is 30s), and the physical layer implements OFDM time-frequency encryption.

[0011] Compared with the prior art, the present invention has the following beneficial effects: This invention proposes an innovative amphibious collaborative control system to solve the switching and collaboration problems of unmanned aerial vehicles in cross-media environments. It uses a deformable duct structure and magnetohydrodynamic thrusters to achieve autonomous reconstruction of the air-water-ground power mode. It also significantly improves the reliability of medium switching and structural durability through a super-hydrophobic coating with a contact angle >150° and a modular shock-absorbing design. It utilizes a federated Kalman filter spatiotemporal alignment and an SVM dynamic decision model to significantly reduce the perception distortion rate to <5% and the switching false trigger rate to 1.3%, generating a high-precision six-dimensional environmental state matrix. The distributed collaborative architecture and medium-adaptive path planning ensure that the multi-machine task response delay is ≤80ms, and achieves a 100% collision avoidance success rate in a Φ28cm underground pipe network. The heterogeneous communication gateway ensures the integrity of cross-media data transmission through a 5G private network with a latency of <10ms and a blue-green laser dual-link redundancy with a rate of 2Mbps. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is a flow chart of the cross-media autonomous switching system of the present invention; Figure 2Schematic diagram of the multi-machine collaborative control architecture of the present invention; Figure 3 A schematic diagram showing a comparison of the path planning algorithms of the present invention; Figure 4 This is a diagram of the heterogeneous communication layered model of the present invention. DETAILED DESCRIPTION

[0013] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.

[0014] A cross-medium low-altitude unmanned aerial vehicle autonomous switching and multi-machine collaborative control system includes a multimodal sensing unit, an autonomous switching unit, and a collaborative control unit, specifically comprising the following steps: S01, according to Figure 1 The cross-media switching control flow chart shown here builds a multimodal perception hardware architecture; S11, the aircraft is equipped with a 77GHz millimeter-wave radar (detection range 0-150m) at the front, a 200kHz multi-beam sonar array (accuracy ±0.05m@5m) in a waterproof compartment (IP68) at the bottom, and a 905nm solid-state LiDAR (angular resolution 0.05°) on the top of the rotating gimbal. S12, the sensor is fixed by a titanium alloy bracket, and the bracket is connected to the body frame with M4 aviation-grade screws. The cables are collected to the main control board through the internal cable management trough, and the signal is transmitted using shielded twisted pair cable (impedance 100Ω); For example, the millimeter-wave radar is mounted on a 15° forward-tilted bracket, the sonar array is arranged in a ring (with a spacing of λ / 2 to prevent interference), and the LiDAR gimbal's pitch adjustment range is ±30°. Each sensor is equipped with a separate PCB adapter board with integrated TVS transient suppression diodes and a 100μH common-mode choke. These are connected to the main control board via a 0.5mm-pitch FFC flexible cable covered with a copper-nickel braided shield.

[0015] S02. Use spatiotemporal synchronous fusion algorithm to process heterogeneous data, establish sensor data mapping matrix, and complete spatiotemporal synchronous fusion processing; S21. Establish sensor timestamp mapping matrix: ; in , , is the position offset, is the installation angle; S22, data alignment is achieved through a sliding window compensation algorithm (window 50ms), with a time domain error of <0.5ms; S03, based on the federated Kalman filter to fuse environmental parameters, construct a six-dimensional feature vector, and generate switching decisions through the SVM classifier; S31. Federal filter fusion environment parameters: Air sub-filter: processes radar data (100Hz) and air pressure data; Water medium sub-filter: Fusion of sonar and fluid pressure data; Subsurface subfilter: integrates LiDAR point cloud (500 points / m²) and IMU data; S32, SVM classifier (γ=0.5) determines the switching condition: ; For example, in tidal water measurements, when the water depth is 1.25 m, the density gradient is 2.3 kg / m³, and the turbidity is 68 NTU for 3 seconds, the SVM output medium probability P = 0.992, generating a switching instruction.

[0016] S04. When the aircraft is crossing the medium, based on the decision instructions of the federated filter and the SVM classifier, the power system structure and energy distribution strategy are changed in real time to perform propulsion system reconstruction; S41, switching from air to underwater: the duct structure closes within 50ms (seal pressure ≥ 12kPa); the magnetohydrodynamic thruster is activated (thrust 5.5N ± 0.3N); the center of gravity adjustment mechanism deflects ΔCG = 3% of the aircraft length; S42, switching from underground to air: deploy the ducted fan (power 850W); activate multi-mode satellite navigation / INS combined positioning (B1C + B2a dual-frequency); release the counterweight (mass 200g); S05. Establish a medium-adaptive collaborative paradigm, achieve triple matching of mission, capability, and environment for heterogeneous UAV swarms through a dynamic contract network protocol, and complete multi-machine collaborative control; S51. Dynamic contract network protocol task allocation: ; S52, medium adaptive path planning: When the system is in air mode, medium adaptive path planning is performed according to the RRT*-Smart algorithm with the constraint condition of turbulence cost factor λ=0.1; when the system is in underwater mode, medium adaptive path planning is performed according to the A*-NS algorithm with the constraint condition of Navier-Stokes fluid integral; when the system is in underground mode, medium adaptive path planning is performed according to the Topo-Dijkstra algorithm with the constraint condition of minimum turning radius R≥0.4m; For example: During an underwater rescue mission, the pilot aircraft (with remaining battery power > 80%) locates the target using sonar, the dynamic contract network protocol allocates three drones to form a triangular search formation within 42ms, and the A-NS algorithm plans a path along the undercurrent (saving 27% energy).

[0017] S06. Build a three-modal intelligent switching heterogeneous communication system and dynamically select the optimal channel through the medium sensing module; S61, main link: 5G private network (20MHz@3.5GHz, end-to-end latency <10ms); S62, backup link: Switching from air to underwater: switching from blue-green laser (532nm, 2Mbps) to underwater acoustic communication (50kHz); Switching from underground to air: using Beidou short messages (transmission delay <5s); S63, Security Encryption System Architecture: A layered encryption strategy is adopted, with differentiated encryption modules deployed on the primary and backup links. These modules are divided into a 5G private network encryption layer, an air-water laser / underwater acoustic encryption layer, and a ground-air BeiDou encryption layer. 5G private network encryption layer: A hardware encryption chip integrating the national secret SM4 algorithm (supporting 20Gbps throughput) and a dynamic key update mechanism based on quantum key distribution (QKD) with a key update cycle of ≤30 seconds is used. The communication module PCB adopts a 16-layer buried resistor design, and key signal traces are wrapped with a serpentine shield. Air-to-water laser / underwater acoustic encryption layer: This layer uses chaotic laser modulation encryption (CLME) technology to achieve waveform obfuscation by injecting disturbances into semiconductor lasers. It also deploys improved OFDM-Hadamard transform encryption, inserts a 32-bit dynamic checksum into each frame of data, and uses an FPGA encryption module (Xilinx Zynq UltraScale+, operating at 400 MHz) in front of the underwater acoustic transducer. Ground-to-air BeiDou encryption layer: Utilizes the BeiDou-3 military-grade encrypted communication protocol (BD-SAM), employs a dual-security structure with 128-bit AES-CBC encryption and CRC32 checksum verification, and integrates a self-destruct fuse circuit (trigger current ≥ 3A) in the SIM card slot. S64, Cross-media Encryption Synchronization Mechanism: Establish a key management unit based on the Trusted Execution Environment (TEE), store the root key through the Hardware Security Module (HSM), trigger the key synchronization protocol when the medium switches, the synchronization delay is ≤50ms, and the key error rate is <10 -9 ; For example, in the Bohai cross-media test, when the aircraft switched from the air (5G link) to the underwater (laser → underwater acoustic link), the encryption controller detected the medium switching instruction (SVM output P = 0.992), triggered the CLME chaotic laser encryption initialization, generated the initial vector IV = 0x7A3E9D (transmitted through the underwater acoustic channel), and loaded the 256-bit dynamic key (update period 2 seconds) on the underwater acoustic receiver FPGA. The measured data transmission bit error rate was ≤10 -6 , successfully resisting 5×10 5 brute force attacks.

[0018] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. The background technology related content presented in this article is intended to provide necessary preparation for understanding the overall concept and innovative context of the present invention. Its purpose is not to confirm, infer or imply that this information has been generally recognized and has become the prior art content mastered by ordinary technicians in this field. Technicians in this industry should understand that the present invention is not limited to the above-mentioned embodiments. The above-mentioned embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the invention to be protected. The scope of protection claimed in the present invention is defined by the attached claims and their equivalents.

Claims

1. A cross-medium low-altitude unmanned aerial vehicle autonomous switching and multi-machine collaborative control system, characterized by: It includes a multimodal perception unit, an autonomous switching unit, and a collaborative control unit, and specifically includes the following steps: S01. Build a multimodal perception unit, deploying a 77GHz millimeter-wave radar, a 200kHz multi-beam sonar, and a 905nm solid-state LiDAR in the main body of the aircraft. The sensors are fixed with titanium alloy brackets, and the sensor cables are connected to the main control board through the cable management trough inside the cabin. S02. Use spatiotemporal synchronous fusion algorithm to process heterogeneous data and establish sensor data mapping matrix: ； The delay compensation , ; S03. Based on the federated Kalman filter, environmental parameters are integrated to construct a six-dimensional feature vector: water depth, density gradient, turbidity, wind speed, methane concentration, and satellite navigation carrier-to-noise ratio. The switching decision is generated through the SVM classifier. When the medium probability P ≥ 0.98 and meets the following conditions: When triggering mode conversion; S04. Execute propulsion system reconstruction: Switching from air to underwater: the duct structure is closed within 50ms and the magnetohydrodynamic thruster is activated; Switching from underground to air: deploy the ducted fans and activate multi-mode satellite navigation / INS combined positioning; S05. Use the improved contract network protocol to achieve multi-machine collaboration, including dynamic link management of heterogeneous communication systems, and the task allocation response time t satisfies: ； Where d is the mission distance, v_max is the maximum speed of the UAV; S06, driving medium adaptive path planner: Air mode: RRT*-Smart algorithm integrated with turbulence cost model; Underwater mode: The A*-NS algorithm is embedded in the Navier-Stokes equation solver, and the cost function is: ； Underground mode: The Topo-Dijkstra algorithm is limited by the pipe network topology constraints.

2. The cross-medium low-altitude unmanned aerial vehicle autonomous switching and multi-machine collaborative control system according to claim 1, characterized in that: The anti-interference hardware architecture design of the multimodal perception unit includes: The millimeter-wave radar is installed at the front of the aircraft, the sonar array is arranged in the bottom waterproof cabin, and the LiDAR is placed on the rotating gimbal at the top; The magnetohydrodynamic propulsion system and the duct structure adopt a modular design and are connected through a quick-release interface. The duct surface is covered with a super-hydrophobic coating. The main control board integrates a three-layer electromagnetic shield, the sensor cable uses shielded twisted pair cables, and the interface uses gold-plated shielded connectors; Each sensor is equipped with 4 silicone shock-absorbing pads, which are installed in a rectangular shape.

3. The cross-medium low-altitude unmanned aerial vehicle autonomous switching and multi-machine collaborative control system according to claim 2, characterized in that: The dynamic switching decision method of the autonomous switching unit includes: S31, the federated filter sets three sub-filters: Air sub-filter: processes millimeter-wave radar data and air pressure data; Water medium sub-filter: fuses sonar data and fluid pressure data; Subsurface subfilter: integrates LiDAR point cloud and IMU data; S32. The main filter of the federated filter realizes cross-domain fusion through the improved covariance crossover algorithm, and outputs a state vector containing the medium type probability P_m∈[0,1] and the interface dynamic index D_i∈R⁺; The S33 and SVM classifiers use the RBF kernel function, and the training data set contains three types of medium boundary samples: tidal waters, turbulent air, and underground pipe networks.

4. The cross-medium low-altitude unmanned aerial vehicle autonomous switching and multi-machine collaborative control system according to claim 1, characterized in that: The multi-machine collaborative control method of the collaborative control unit includes: S41. Dynamic role allocation mechanism based on comprehensive scoring function: ； Where Q is the remaining power, D is the relative target distance, is the sensor resolution; S42, distributed fault-tolerant protocol execution process: Fault detection: If the heartbeat packet timeout is greater than 100ms, the node is considered to be faulty; Task redistribution: completed within 200ms through the Paxos consensus algorithm; Path replanning: RRT*-Smart algorithm convergence time < 500ms.

5. The cross-medium low-altitude unmanned aerial vehicle autonomous switching and multi-machine collaborative control system according to claim 1, characterized in that: The heterogeneous communication system includes: S51, the main link uses 5G private network, and the end-to-end delay is ≤10ms; S52, the backup link uses a combination of blue-green laser communication and underwater acoustic communication; S53, communication protocol conversion process: Switching from air to underwater: interrupt the 5G link and activate the underwater acoustic communication module; Switch from underwater to air: turn off the sonar modulator and activate the laser transmitter; S54, the data security layer adopts SM4 national encryption algorithm for encryption, and the physical layer implements OFDM time-frequency encryption.

Images