Unmanned aerial vehicle on-board 5G radar low-altitude intelligent sensing system

Abstract

The application discloses a kind of unmanned aerial vehicle airborne 5G radar low altitude wisdom perception system, the system includes unmanned aerial vehicle group and its carrying flight control module, millimeter wave radar detection module, airborne computing processing module, obstacle avoidance module, 5G communication module and positioning module, wherein radar-time space weighting algorithm is embedded in obstacle avoidance module, can real-time fusion millimeter wave radar three-dimensional point cloud and centimeter level positioning information of 5G sensing integrated base station, complete local path re-planning in short time, and realize end-to-end control closed loop through 5G-AuRLLC slice.The system automatically switches airborne offline decision when network interruption, and synchronizes global risk map after link recovery.The success rate of obstacle avoidance is higher than traditional technology in the actual measurement of complex obstacle scene in urban low altitude, and the energy consumption is lower than traditional scheme.The application effectively solves the problems of large perception delay, redundant obstacle avoidance path and poor link security in existing unmanned aerial vehicle low altitude operation, and can be widely applied in urban inspection, emergency delivery, low altitude logistics and other scenes.

Description

Technical Field

[0001] This invention relates to the technical field of low-altitude flight of unmanned aerial vehicles (UAVs), specifically to an airborne 5G radar low-altitude intelligent sensing system for UAVs. Background Technology

[0002] Inspection drones are deployed in the field environment and are mainly used to inspect the work site when no one is on duty. This can save commuting costs. When conducting low-altitude inspections, inspection drones may encounter obstacles such as other drones, kites, balloons, large aircraft taking off or landing, and birds.

[0003] Currently, there is extensive research on obstacle avoidance for drones both domestically and internationally, generally relying on 3D elevation maps, binocular cameras, and other high-precision equipment. When drones fly at low altitudes, further demands are placed on their inspection range detection and emergency response capabilities. However, existing technologies have the following shortcomings:

[0004] Insufficient perception dimensions: Traditional visual / laser SLAM is greatly affected by lighting and rain / fog, and its update frequency is only 10-20Hz, making it unable to capture high-speed dynamic obstacles in real time.

[0005] Communication-perception disconnect: The separation of radar and communication links leads to spectrum conflicts and link delays >100ms, making it difficult to meet millisecond-level obstacle avoidance requirements.

[0006] Inefficient path planning: Classic obstacle avoidance algorithms only consider static obstacles, have a single heuristic function, and are prone to sharp turns and flying around obstacles, increasing energy consumption and collision risk.

[0007] Poor network robustness: Once the 5G link is lost, the drone must return to base or hover immediately, resulting in a high mission interruption rate.

[0008] Therefore, there is an urgent need for an unmanned aerial vehicle (UAV) system that can achieve integrated "high-frequency perception, low-latency decision-making, and high-reliability execution" in complex low-altitude environments. Summary of the Invention

[0009] The purpose of this invention is to provide an airborne 5G radar low-altitude intelligent sensing system for unmanned aerial vehicles (UAVs) to solve the problems mentioned in the background art.

[0010] The technical solution adopted in this invention is: to provide an unmanned aerial vehicle (UAV) airborne 5G radar low-altitude intelligent sensing system, comprising:

[0011] The unmanned aerial vehicle (UAV) group is equipped with a flight control module, a millimeter-wave radar detection module, an airborne computing and processing module, an obstacle avoidance module, a 5G communication module, and a positioning module.

[0012] The flight control module is used to control the drone's flight according to a preset low-altitude flight route and to adjust the drone's flight trajectory according to the obstacle avoidance command.

[0013] The millimeter-wave radar detection module is mounted on the underside of the UAV and operates in an electronic scanning manner to acquire three-dimensional point cloud data of the low-altitude environment and send the three-dimensional point cloud data to the airborne computing and processing module.

[0014] The airborne computing and processing module is connected to the 5G communication module and the millimeter-wave radar module, and is used to process the three-dimensional point cloud data in real time.

[0015] The 5G communication module is used for low-latency two-way communication with the remote control center.

[0016] The obstacle avoidance module is connected to the onboard computing and processing module and is used to generate obstacle avoidance commands based on the processed 3D point cloud data; the obstacle avoidance module further includes:

[0017] The dynamic threat classification unit is used to classify obstacles detected in the three-dimensional point cloud data into multiple threat levels according to their relative speed, distance, and geometric dimensions.

[0018] The path replanning unit is used to generate local obstacle avoidance paths in real time based on the multi-level threat levels.

[0019] The priority arbitration unit is used to select the priority path that consumes the least energy and meets the safety margin threshold when multiple obstacle avoidance paths exist simultaneously.

[0020] Preferably, the positioning module adopts the BeiDou satellite positioning system, which acquires high-precision position information of the inspection drone in real time and stores the position information in association with three-dimensional point cloud data in order to accurately determine the location of obstacles and anomalies.

[0021] Preferably, the radar module employs frequency-modulated continuous wave radar technology, which continuously transmits electromagnetic wave signals whose frequency changes linearly with time, and determines the target's distance and speed information by analyzing the frequency difference between the received signal and the transmitted signal.

[0022] Preferably, when the dynamic threat classification unit detects the presence of an obstacle, it constructs an obstacle hazard assessment model to predict the threat level of the obstacle to the UAV. The obstacle hazard assessment model is defined as follows:

[0023] ;

[0024] in, This indicates the level of danger that obstacle i poses to the drone. The danger level is determined by a preset threat level range, which includes high risk, medium risk, and low risk. represents the magnification factor, reflecting the amplification effect of relative movement and obstacle area on the degree of danger; v represents the relative speed of obstacle i; and d represents the relative distance between obstacle i and the drone. Represents the cross-sectional area of ​​obstacle i. Indicates the safe avoidance area. This represents an adjustment factor, used to indicate the degree of influence of obstacle avoidance reaction time on the obstacle hazard level. Indicates the drone's avoidance reaction time. This represents the acceleration reward intensity coefficient, where This indicates the threat level posed to the drone when its acceleration exceeds a safety threshold. Indicates the acceleration of the drone. This indicates the adjustment coefficient for the avoidance area.

[0025] Preferably, the path replanning unit adopts a method for low-altitude UAVs to perceive the scene using millimeter-wave radar. It uses heuristic functions, node expansion rules, cost functions, and termination conditions to convert the real-time 3D point cloud data and velocity field information generated by the millimeter-wave radar into a local obstacle avoidance path that the UAV can execute immediately.

[0026] More preferably, the heuristic function is as follows:

[0027] h(n) = α·Δt + β·E + γ·σ;

[0028] Where h(n) represents the heuristic function, Δt is the time increment, E is the energy consumption increment, σ is the safety margin deviation, and α, β and γ are weighting coefficients and α+β+γ=1.

[0029] Preferably, the airborne computing processing module includes a signal preprocessing unit, a target detection and tracking unit, and a feature extraction and analysis unit. The signal preprocessing unit is used to perform preprocessing operations such as filtering, amplification, and digitization on the radar echo signal. The target detection and tracking unit detects and tracks low-altitude targets in real time based on the preprocessed signal using target detection and tracking algorithms. The feature extraction and analysis unit extracts features from the targets detected by the millimeter-wave radar detection module. The features include the size, shape, and trajectory of the target.

[0030] Preferably, the 5G communication module supports multiple 5G frequency bands and communication standards, selects the optimal 5G frequency band and communication mode for data transmission based on the actual communication environment and business needs during flight, and has data encryption functionality.

[0031] By adopting the above technical solutions, the safety of drones during low-altitude flight is ensured.

[0032] Compared with existing technologies, the beneficial effects of this invention are as follows: The UAV-borne 5G radar low-altitude intelligent perception system provided by this invention achieves end-to-end low latency through a radar-spatiotemporal weighted algorithm embedded in the obstacle avoidance module; the fusion of centimeter-level positioning and millimeter-level radar greatly reduces positioning errors, enabling centimeter-level safety margin flight in dense obstacle areas; the heuristic function simultaneously incorporates time, energy consumption, and safety margin weights, effectively extending the measured flight endurance and reducing the number of sudden braking and sharp turns; offline obstacle avoidance can still be performed within 30 seconds of network interruption, and global map synchronization is completed within 500ms after link recovery, significantly improving mission continuity; in addition, this invention is compatible with multiple UAV platforms and 5G-A base stations, requires no additional LiDAR, has low hardware costs, and is easy to deploy on a large scale. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the structure of an airborne 5G radar low-altitude intelligent sensing system for unmanned aerial vehicles (UAVs) according to the present invention.

[0034] Figure 2 This is a schematic diagram of the structural relationship of an airborne 5G radar low-altitude intelligent sensing system for unmanned aerial vehicles (UAVs) according to the present invention.

[0035] Figure 3 This is a schematic diagram of the obstacle avoidance module structure of the present invention. Detailed Implementation

[0036] 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.

[0037] Please see Figure 1-3 A drone-borne 5G radar low-altitude intelligent sensing system is provided, comprising:

[0038] The unmanned aerial vehicle (UAV) group is equipped with a flight control module, a millimeter-wave radar detection module, an airborne computing and processing module, an obstacle avoidance module, a 5G communication module, and a positioning module.

[0039] The flight control module is used to control the drone's flight according to a preset low-altitude flight route and to adjust the drone's flight trajectory according to the obstacle avoidance command.

[0040] The millimeter-wave radar detection module is mounted on the underside of the UAV and operates in an electronic scanning manner to acquire three-dimensional point cloud data of the low-altitude environment and send the three-dimensional point cloud data to the airborne computing and processing module.

[0041] The airborne computing and processing module is connected to the 5G communication module and the millimeter-wave radar module, and is used to process the three-dimensional point cloud data in real time.

[0042] The 5G communication module is used for low-latency two-way communication with the remote control center.

[0043] The obstacle avoidance module is connected to the onboard computing and processing module and is used to generate obstacle avoidance commands based on the processed 3D point cloud data; the obstacle avoidance module further includes:

[0044] The dynamic threat classification unit is used to classify obstacles detected in the three-dimensional point cloud data into multiple threat levels according to their relative speed, distance, and geometric dimensions.

[0045] The path replanning unit is used to generate local obstacle avoidance paths in real time based on the multi-level threat levels.

[0046] The priority arbitration unit is used to select the priority path that consumes the least energy and meets the safety margin threshold when multiple obstacle avoidance paths exist simultaneously.

[0047] In this embodiment, the 5G communication module selected is the Huawei MH5000 5G communication module, which supports 5G NR standalone networking mode and has uRLLC slicing function. It is installed in the signal compartment at the rear of the UAV and connected to the airborne antenna via a Fakra interface. The antenna uses a high-gain Yagi antenna to ensure stable communication with the 5G-A base station in low-altitude environments. This module can achieve data transmission rates of up to 2Gbps with end-to-end latency of less than 10ms.

[0048] The 5G communication module connects to the onboard computing and processing module via a PCIe interface (or other high-speed communication interface). The 5G communication module is responsible for communicating with the ground control center or 5G base station, uploading the UAV's status data (such as position, speed, and battery level), and receiving control commands or updated mission information from the ground. Simultaneously, the 5G communication module provides the UAV with centimeter-level high-precision positioning data (via a 5G-A integrated sensing base station). This data is transmitted to the onboard computing and processing module to assist in path planning and obstacle avoidance decisions. Furthermore, during collaborative operations, the 5G communication module is also used to interact with other UAVs, enabling multi-UAV collaborative obstacle avoidance.

[0049] The millimeter-wave radar detection module uses Infineon's IRC1020 radar chip, integrating a 4-transmit, 4-receive antenna array. It is installed in a radar pod beneath the UAV fuselage and connected to the onboard computing module via a coaxial cable. The radar has 128 virtual receiving channels, a range resolution better than 3cm, and is capable of real-time perception of the low-altitude environment using electronic scanning.

[0050] The millimeter-wave radar detection module is connected to the airborne computing and processing module via a coaxial cable (or high-speed data bus). The millimeter-wave radar scans the surrounding environment in real time (including the position, speed, and shape of obstacles) and transmits this data to the airborne computing and processing module in the form of point clouds. After receiving the radar data, the airborne computing and processing module processes and analyzes it using a "radar-spatiotemporal weighted" algorithm to generate obstacle avoidance commands.

[0051] The airborne computing processing module is built on NVIDIA Jetson OrinNX and features a heterogeneous computing architecture, including an FPGA daughter card for radar signal processing, a GPU daughter card for deep learning inference, and a 6-core ARM CPU for task scheduling. It is installed in the main control cabin inside the UAV fuselage and is connected to the 5G communication module and millimeter-wave radar module via a PCIe interface, enabling real-time processing of radar data and efficient operation of obstacle avoidance algorithms.

[0052] Obstacle avoidance module: As a software functional unit of the airborne computing and processing module, it runs a self-developed radar-temporal weighted algorithm, which can receive radar point cloud data in real time and generate obstacle avoidance commands; The flight control module is developed and customized using DJI's FlightController SDK, and is installed in the flight control cabin at the front of the UAV fuselage. It is connected to the obstacle avoidance module via a CAN bus, and can receive obstacle avoidance commands and accurately adjust the UAV's flight attitude and trajectory.

[0053] The obstacle avoidance module is actually a software functional unit within the airborne computing and processing module, running on the module's processor. It receives obstacle information from the millimeter-wave radar detection module, combines it with high-precision positioning data and other relevant information provided by the 5G communication module, runs a "radar-spatiotemporal weighted" algorithm to perform path planning and obstacle avoidance decisions, and sends the generated obstacle avoidance commands to the flight control module to control the drone to avoid obstacles.

[0054] The flight control module is connected to the onboard computing and processing module via a high-speed CAN bus (or UART interface). It is responsible for receiving control commands sent by the onboard computing and processing module and converting them into specific flight actions, such as adjusting flight attitude (pitch, roll, yaw), changing flight speed, and altitude. At the same time, the flight control module provides real-time feedback on the UAV's current flight status (such as position, speed, attitude, etc.) to the onboard computing and processing module, so that the onboard computing and processing module can make path planning and obstacle avoidance decisions based on the actual situation.

[0055] The positioning module uses the BeiDou satellite positioning system to acquire high-precision location information of the inspection drone in real time, and associates and stores the location information with three-dimensional point cloud data in order to accurately determine the location of obstacles and anomalies.

[0056] The positioning module (such as a GPS / RTK module) connects to the onboard computing module via a UART interface (or other universal serial interface). The positioning module receives satellite signals, calculates the UAV's precise location information (including latitude, longitude, and altitude) in real time, and sends it to the onboard computing module. The onboard computing module combines this location information with obstacle data from the millimeter-wave radar detection module to more accurately determine the UAV's position in space, thereby improving the reliability and accuracy of obstacle avoidance.

[0057] During actual flight of the drone, the various modules work closely together. The millimeter-wave radar detection module monitors the surrounding environment in real time and provides obstacle data; the positioning module provides the drone's precise location information; the 5G communication module is responsible for data communication and positioning enhancement with external systems; the onboard computing and processing module receives this data, runs obstacle avoidance algorithms for path planning and decision-making; finally, the flight control module, according to the instructions from the onboard computing and processing module, precisely controls the drone's flight attitude and trajectory to achieve autonomous obstacle avoidance and safe flight.

[0058] The dynamic threat classification unit classifies obstacles detected by radar according to their relative speed, distance, and geometric dimensions, dividing the threat level into three levels: T1 (low threat), T2 (medium threat), and T3 (high threat), providing a decision-making basis for obstacle avoidance algorithms.

[0059] When an obstacle is detected, an obstacle hazard assessment model is constructed to predict the threat level of the obstacle to the drone. The obstacle hazard assessment model is defined as follows:

[0060] ;

[0061] in, This indicates the level of danger that obstacle i poses to the drone. The danger level is determined by a preset threat level range, which includes high risk, medium risk, and low risk. represents the magnification factor, reflecting the amplification effect of relative movement and obstacle area on the degree of danger; v represents the relative speed of obstacle i; and d represents the relative distance between obstacle i and the drone. Represents the cross-sectional area of ​​obstacle i. Indicates the safe avoidance area. This represents an adjustment factor, used to indicate the degree of influence of obstacle avoidance reaction time on the obstacle hazard level. Indicates the drone's avoidance reaction time. This represents the acceleration reward intensity coefficient, where This indicates the threat level posed to the drone when its acceleration exceeds a safety threshold. Indicates the acceleration of the drone. This indicates the adjustment coefficient for the avoidance area.

[0062] The path replanning unit constructs feasible obstacle avoidance paths in real time within a local space based on threat level, prioritizing the path with the lowest energy consumption and meeting the safety margin threshold, thus ensuring the efficiency and safety of the UAV during obstacle avoidance.

[0063] This paper adopts a millimeter-wave radar perception scenario for low-altitude UAVs, and transforms the real-time 3D point cloud data and velocity field information generated by the millimeter-wave radar into a local obstacle avoidance path that the UAV can execute immediately through heuristic functions, node expansion rules, cost functions and termination conditions.

[0064] The radar-spatiotemporal weighted algorithm of this invention is specifically optimized for millimeter-wave radar perception scenarios of low-altitude UAVs. The algorithm focuses on four key aspects: heuristic function, node expansion rules, cost function, and termination condition. The core of the algorithm is as follows:

[0065] State-space representation:

[0066] The coordinates of node n are (x, y, z, t), where z is the height layer and t is the discrete timestamp; each edge e carries both spatial cost Δs and energy cost ΔE.

[0067] The heuristic function is as follows:

[0068] h(n) = α·Δt + β·E + γ·σ;

[0069] Where h(n) represents the heuristic function, Δt is the time increment, E is the energy consumption increment, σ is the safety margin deviation, and α, β and γ are weighting coefficients and α+β+γ=1.

[0070] In node n, the heuristic function is as follows:

[0071] h(n)=α·Δt(n)+β·En+γ·σn;

[0072] Where Δt(n) represents the shortest time estimate from node n to the target considering the maximum airspeed limit; En represents the estimated energy consumption from node n to the target, calculated by the battery model and the headwind factor; σn represents the safety margin deviation, which is equal to the 3σ confidence interval distance from node n to the nearest radar detection point.

[0073] The algorithm can process 3D point cloud data acquired by millimeter-wave radar in real time, and combine it with centimeter-level positioning information provided by 5G-A base stations to complete local path replanning within 30ms, and generate obstacle avoidance commands to send to the flight control module to realize the drone's autonomous obstacle avoidance.

[0074] For each cluster of dynamic point clouds returned by millimeter-wave radar, its predicted position after time Δt is calculated, generating a spatiotemporal occupancy grid (4D-OccupancyGrid). If the cost of node n falls within a future occupancy grid, an additional penalty term is added.

[0075] ;

[0076] Where P represents an additional penalty term, and k is an adjustable coefficient. This represents the relative speed of obstacle i. Indicates the closest distance.

[0077] Priority arbitration unit: When multiple obstacle avoidance paths exist simultaneously, the optimal path is selected for execution by comprehensively considering factors such as path safety, energy consumption, and flight efficiency.

[0078] Add two constraints when expanding child nodes:

[0079] • Maneuvering constraints: Roll angle and climb rate must not exceed the physical limits of the UAV;

[0080] • Communication constraints: Child nodes must fall within the coverage area where the current 5G signal quality is >100dBm; nodes that do not meet any of the constraints will be pruned.

[0081] Early termination condition: The path can be returned early when the first node n_goal that satisfies the following formula appears in the Open list:

[0082] g(n_goal)+h(n_goal)≤(1+ε)·g_best;

[0083] And σ_safe(n_goal)≥σ_min, where ε is an approximation factor (typical value 0.1) and σ_min is the preset minimum safe distance threshold.

[0084] Local replanning trigger:

[0085] If the radar detects a new high-threat obstacle (threat level ≥ T3) within L meters ahead of the planned path, the improved algorithm is immediately rerun from the current position. L and T3 are given in real time by the dynamic threat classification unit.

[0086] With the above improvements, the algorithm can complete local replanning in a 100m×100m×20m airspace in less than 30ms on the onboard GPU while keeping the UAV on the optimal path, meeting the requirements of 5GuRLLC scenarios with 10ms end-to-end latency.

[0087] In a 5G communication module environment, it exhibits the following key advantages and real-world performance:

[0088] 1. Latency performance: Meets millisecond-level obstacle avoidance closed-loop requirements;

[0089] Communication latency: The 5G-A integrated sensing base station achieves an end-to-end latency of <10ms (actual measurement). Combined with the local replanning time of the radar-A* algorithm of <30ms, the overall control closed-loop latency is less than 40ms.

[0090] Edge collaboration: Based on 5G-A edge computing nodes, the radar algorithm can offload some of the node's extended computing, reducing the local GPU load by about 35% in actual tests, and further shortening the replanning time to less than 20ms.

[0091] 2. Sensing update frequency: 100Hz dynamic environment response;

[0092] The 5G integrated sensing base station supports 100 airspace scans per second (i.e., a 10ms update cycle), and the radar-A* algorithm can synchronously acquire the latest obstacle point cloud. The dynamic threat classification unit refreshes the threat level in real time, avoiding the collision risk caused by "perception lag".

[0093] 3. Path planning accuracy: centimeter-level positioning + millimeter-level radar;

[0094] Positioning accuracy: 5G network provides centimeter-level RTK positioning (error <10cm), combined with millimeter-wave radar with a distance resolution of 3cm, radar algorithm can achieve centimeter-level obstacle avoidance accuracy in local path planning.

[0095] Real-world test case: In the Shenzhen Huawei Bantian 5.5GPark test, the drone equipped with radar algorithm achieved a dynamic obstacle avoidance success rate of >98% in areas densely packed with obstacles (such as tree canopies and lampposts) with a diameter of 0.3m, and the path smoothness was improved by 50%.

[0096] 4. Network robustness: Autonomous decision-making during connection failure + recovery and synchronization;

[0097] Disconnection scenario: When the 5G-A signal is interrupted due to obstruction or interference, the radar-A* algorithm switches to airborne offline mode and continues obstacle avoidance by relying on the local radar point cloud (the safety boundary is automatically expanded by 50%). In actual tests, there were no collision events within 30 seconds of the interruption.

[0098] Recovery mechanism: After the link is restored, the algorithm synchronizes the global risk map through the 5G-A network, and the path deviation during the link loss is corrected within 500ms.

[0099] 5. Resource consumption: Integrated communication and sensing reduce spectrum conflicts;

[0100] Spectrum efficiency: The 5G integrated sensing base station adopts the 4.9GHz shared frequency band. The radar-temporal weighted algorithm does not need to monopolize the radar spectrum. The measured communication rate only decreases by 5%, which is far lower than the traditional radar-communication separation scheme (decrease of 20%+).

[0101] Energy efficiency optimization: Thanks to the low power consumption design of 5G-A (such as sleep mode), the total power consumption of the drone system equipped with the radar-A* algorithm is <15W, and the flight time is extended by 12%-15%.

[0102] In summary, the radar-spatiotemporal weighted algorithm achieves closed-loop obstacle avoidance with "low latency, high refresh rate, and high accuracy" in a 5G environment. It is particularly suitable for scenarios such as urban low-altitude areas, complex obstacles, and dynamic threats, and is ready for large-scale commercial use. The following is the working principle of this invention:

[0103] Before takeoff, flight mission parameters (such as start point, destination, flight altitude, and flight speed) are input into the airborne computing and processing module. Simultaneously, weighting coefficients α, β, and γ in the radar-spatiotemporal weighted algorithm are set according to mission requirements. The airborne computing and processing module performs preliminary path planning based on the mission parameters and digital map, and stores the planning results in memory.

[0104] After takeoff, the millimeter-wave radar module continuously scans the surrounding environment at a frequency of 100Hz, acquiring real-time 3D point cloud data and transmitting it to the onboard computing module. Simultaneously, the 5G-A communication module receives centimeter-level positioning information from the base station, ensuring the drone accurately knows its location. The onboard computing module processes the radar data in real time, the dynamic threat grading unit classifies obstacles into threat levels, and the radar-spatiotemporal weighted A* algorithm, based on the current environment and location information combined with the threat level, completes local path replanning within 30ms and sends obstacle avoidance commands to the flight control module. The flight control module adjusts the drone's flight attitude and trajectory in real time according to the obstacle avoidance commands, ensuring the drone avoids obstacles and flies along the planned safe path.

[0105] Furthermore, when multiple drones collaborate on missions, each drone shares its flight status and perceived environmental information in real time via a 5G-A network. Each drone's onboard computing module incorporates this information into a radar-spatiotemporal weighted algorithm to achieve collaborative obstacle avoidance. The algorithm rationally plans its own flight trajectory based on the position, speed, and flight path of other drones, avoiding collisions and improving the safety and efficiency of collaborative operations.

[0106] This invention also includes a network interruption emergency handling mechanism: when the 5G communication link is interrupted, the obstacle avoidance module automatically switches to an airborne offline decision-making mode. In this mode, the radar-spatiotemporal weighted algorithm performs path planning solely based on local radar data, while the safety boundary is automatically expanded by 50% to cope with unforeseen circumstances. The UAV continues to perform obstacle avoidance operations in offline mode to ensure flight safety. Once the link is restored, the UAV immediately uploads its local perception data to the base station and receives updated global map information, quickly resuming collaborative operations.

[0107] The UAV-borne 5G radar low-altitude intelligent sensing system in this embodiment achieves efficient autonomous obstacle avoidance for UAVs in complex low-altitude environments through the coordinated operation of the 5G-A communication module, millimeter-wave radar module, airborne computing and processing module, obstacle avoidance module, and flight control module, combined with radar-spatiotemporal weighted algorithm. It has advantages such as ultra-low latency, high-precision positioning, dual optimization of energy and safety, and autonomous decision-making in case of link failure. It can be widely used in urban inspection, emergency delivery, low-altitude logistics and other fields.

[0108] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of this invention. In practical applications, those skilled in the art can select some or all of the workflow to achieve the purpose of this embodiment according to actual needs, and no restrictions are imposed here.

[0109] In addition, for technical details not described in detail in this embodiment, please refer to the parameter operation method provided in any embodiment of the present invention, which will not be repeated here.

[0110] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0111] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0112] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A UAV-borne 5G radar low-altitude intelligent sensing system, characterized in that, include: The unmanned aerial vehicle (UAV) group is equipped with a flight control module, a millimeter-wave radar detection module, an airborne computing and processing module, an obstacle avoidance module, a 5G communication module, and a positioning module. The flight control module is used to control the drone's flight according to a preset low-altitude flight route and to adjust the drone's flight trajectory according to the obstacle avoidance command. The millimeter-wave radar detection module is mounted on the underside of the UAV and operates in an electronic scanning manner to acquire three-dimensional point cloud data of the low-altitude environment and send the three-dimensional point cloud data to the airborne computing and processing module. The airborne computing and processing module is connected to the 5G communication module and the millimeter-wave radar module, and is used to process the three-dimensional point cloud data in real time. The 5G communication module is used for low-latency two-way communication with the remote control center. The obstacle avoidance module is connected to the airborne computing and processing module and is used to generate obstacle avoidance commands based on the processed 3D point cloud data. The obstacle avoidance module further includes: The dynamic threat classification unit is used to classify obstacles detected in the three-dimensional point cloud data into multiple threat levels according to their relative speed, distance, and geometric dimensions. The path replanning unit is used to generate local obstacle avoidance paths in real time based on the multi-level threat levels. The priority arbitration unit is used to select the priority path that consumes the least energy and meets the safety margin threshold when multiple obstacle avoidance paths exist simultaneously. When the dynamic threat classification unit detects the presence of an obstacle, it constructs an obstacle hazard assessment model to predict the threat level of the obstacle to the UAV. The obstacle hazard assessment model is defined as follows: ； in, This indicates the level of danger that obstacle i poses to the drone. The danger level is determined by a preset threat level range, which includes high risk, medium risk, and low risk. represents the magnification factor, reflecting the amplification effect of relative movement and obstacle area on the degree of danger; v represents the relative speed of obstacle i; and d represents the relative distance between obstacle i and the drone. Represents the cross-sectional area of ​​obstacle i. Indicates the safe avoidance area. This represents an adjustment factor, used to indicate the degree of influence of obstacle avoidance reaction time on the obstacle hazard level. Indicates the drone's avoidance reaction time. This represents the acceleration reward intensity coefficient, where This indicates the threat level posed to the drone when its acceleration exceeds a safety threshold. Indicates the acceleration of the drone. This indicates the adjustment coefficient for the avoidance area.

2. The UAV-borne 5G radar low-altitude intelligent sensing system according to claim 1, characterized in that, The positioning module uses the BeiDou satellite positioning system to acquire high-precision location information of the inspection drone in real time, and stores the location information in association with three-dimensional point cloud data in order to accurately determine the location of obstacles and anomalies.

3. The UAV-borne 5G radar low-altitude intelligent sensing system according to claim 1, characterized in that, The radar module employs frequency-modulated continuous wave radar technology, which continuously transmits electromagnetic wave signals whose frequency varies linearly with time, and determines the target's distance and speed information by analyzing the frequency difference between the received and transmitted signals.

4. The UAV-borne 5G radar low-altitude intelligent sensing system according to claim 1, characterized in that, The path replanning unit adopts a millimeter-wave radar perception scenario for low-altitude UAVs. It uses heuristic functions, node expansion rules, cost functions, and termination conditions to convert the real-time 3D point cloud data and velocity field information generated by the millimeter-wave radar into a local obstacle avoidance path that the UAV can execute immediately.

5. The UAV-borne 5G radar low-altitude intelligent sensing system according to claim 4, characterized in that, The heuristic function is as follows: h(n) = α·Δt + β·E + γ·σ; Where h(n) represents the heuristic function, Δt is the time increment, E is the energy consumption increment, σ is the safety margin deviation, and α, β and γ are weighting coefficients and α+β+γ=1.

6. The UAV-borne 5G radar low-altitude intelligent sensing system according to claim 1, characterized in that, The airborne computing and processing module includes a signal preprocessing unit, a target detection and tracking unit, and a feature extraction and analysis unit. The signal preprocessing unit performs filtering, amplification, and digitization preprocessing operations on the radar echo signal. The target detection and tracking unit detects and tracks low-altitude targets in real time based on the preprocessed signal using target detection and tracking algorithms. The feature extraction and analysis unit extracts features from the targets detected by the millimeter-wave radar detection module, including the target's size, shape, and trajectory.

7. The UAV-borne 5G radar low-altitude intelligent sensing system according to claim 1, characterized in that, The 5G communication module supports multiple 5G frequency bands and communication standards. It selects the optimal 5G frequency band and communication mode for data transmission based on the actual communication environment and business needs during flight, and also has data encryption capabilities.

Images