Unmanned aerial vehicle operation ground risk mitigation method, device and system, and storage medium

Abstract

The invention relates to the technical field of unmanned aerial vehicles, in particular to an unmanned aerial vehicle operation ground risk relieving method, device and system and a storage medium, and the method comprises the following steps: S1, obtaining a fault state of an unmanned aerial vehicle, and predicting a falling area and an impact probability based on a trajectory descent model; s2, quantifying a regional risk value, and generating a risk distribution diagram by fusing the collision probability; s3, performing global path planning by adopting an A * algorithm based on dynamic weight and adaptive step length according to an assessment result of the risk distribution map; and S4, evaluating the path point risk in real time in the flight process, and if the path point risk exceeds the dynamic threshold value, executing local path re-planning and adjusting the flight path based on a rapid expansion random tree algorithm of risk perception and increment adjustment. According to the invention, through fusion of the global path planning based on risk distribution and the rapid expansion random tree algorithm based on risk perception and increment adjustment, accurate avoidance of the unmanned aerial vehicle to the ground risk in the dynamic environment and intelligent optimization of the flight path are realized.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) technology, and in particular to methods, devices, systems, and storage media for mitigating ground-related risks during UAV operation. Background Technology

[0002] With the large-scale development of the low-altitude economy, the application scenarios of drones in civilian fields such as logistics delivery, agricultural plant protection, and emergency rescue are becoming increasingly complex. During missions, drones, due to their low flight altitude, highly dynamic flight paths, and significant susceptibility to environmental interference, are prone to flight control failures and collisions, leading to frequent crashes and posing a serious threat to ground personnel, facilities, and the ecological environment. Before drones are widely deployed, it is urgent to develop systematic strategies and methods for mitigating the ground risks associated with drone operations to ensure urban low-altitude safety and support sustainable industrial development.

[0003] Existing risk mitigation measures in drone operation management can be mainly summarized into the following three categories: (1) Risk mitigation based on speed adjustment: The speed of the UAV is dynamically adjusted according to the risk assessment results. In low-risk areas, the normal flight speed is maintained, and in high-risk areas, the emergency response time is extended by slowing down or the exposure time of risk is reduced by speeding up, thereby reducing the impact of the accident.

[0004] (2) Risk mitigation measures for setting up safe zones: By using technologies such as electronic fences to delineate safe flight zones and restrict drones from flying in densely populated areas and around critical infrastructure, potential risks can be effectively reduced, but airspace resource utilization will be significantly reduced.

[0005] (3) Risk mitigation measures based on route planning: Combine the real-time risk assessment results, use the route planning algorithm to optimize the flight path, avoid dangerous areas and adjust the route to avoid collisions, and at the same time try to avoid areas with human activity, thereby reducing the threat to ground targets.

[0006] Current technological systems still have the following limitations in the research of risk assessment and mitigation for unmanned aerial vehicle (UAV) operations, making it difficult to meet high-level safety requirements. These limitations are specifically manifested in the following aspects: (1) Some mitigation measures lack flexibility: For example, the safe zone restriction method adopts a static airspace division mode, which is difficult to adapt to the dynamic demand for airspace resources in diverse mission scenarios, resulting in an imbalance between airspace utilization and safety objectives.

[0007] (2) Insufficient comprehensive consideration of multiple factors: Existing studies mostly focus on single risk factors and do not fully integrate multi-source data such as meteorological changes and dynamic changes in population density, which leads to a decrease in the effectiveness of risk mitigation measures in complex scenarios.

[0008] (3) Lack of emergency response mechanism: The existing path planning is mainly based on global static planning and has not established a real-time risk response mechanism for emergency interference, making it difficult to avoid the risk of accidental fall during the execution of the preset path.

[0009] (4) Lack of systematic research on risk mitigation measures: Current research often conducts risk assessment and mitigation strategy design in isolation, without forming a closed-loop system of "assessment-planning-adjustment", resulting in the disconnect between mitigation measures and real-time risk situation. Summary of the Invention

[0010] This invention provides a method, apparatus, system, and storage medium for mitigating ground-to-ground risks during unmanned aerial vehicle (UAV) operation, in order to minimize the potential threat to ground targets during UAV flight.

[0011] Methods for mitigating ground-related risks from drone operations include the following steps: S1, obtain the fault status of the UAV during flight, and based on the fault status, predict the crash area and the probability of impact in different areas after the UAV malfunctions by using a UAV crash model based on ballistic descent. S2, quantify the risk value of the area based on the relevant factors affecting the risk level in the area, and weight and integrate the impact probability of different fall areas with the risk value in the area to generate a risk distribution map; S3. Based on the quantitative assessment of UAV ground risk in the risk distribution map, the A* algorithm based on dynamic weights and adaptive step size is used to plan the global flight path of the UAV. S4. During the operation of the UAV along the initial global operating path, the risk value in the dynamic environment is evaluated in real time based on the risk distribution map. When the risk value at the path point is detected to be no less than the dynamic risk threshold, the online path local optimization mechanism is activated. The risk perception cost function and dynamic step size control are constructed through the fast extended random tree algorithm based on risk perception and incremental adjustment. The flight path from the current point to the target location is recalculated and the UAV flies accordingly.

[0012] Optionally, S1 includes: S11 determines the fault status of the UAV by collecting factors that affect its reliability during operation, including positioning error identification, weather disturbance identification, system and equipment reliability assessment, operator factor analysis, and flight airspace and mode determination. The positioning error identification and analysis are caused by navigation errors due to GPS ephemeris errors, clock errors, signal propagation errors, and geometric configuration errors. The meteorological disturbance identification and assessment method evaluates the impact of environmental factors such as wind speed and direction, precipitation, haze, temperature and air pressure changes on the flight stability of the UAV. The system and equipment reliability assessment includes identifying abnormal conditions such as battery power, flight control failure, communication interruption, and aging components. The operator factor analysis considers flight anomalies caused by human factors such as lack of experience, misoperation, and slow response. The determination of flight airspace and mode is based on risk level classification according to whether it is low-altitude, ultra-low-altitude, VLOS or BVLOS mode. S12. Using the UAV ballistic descent mode and combined with modeling analysis, the falling motion of the UAV after a malfunction is analyzed. By establishing the force and motion equations in the horizontal and vertical directions, and expressing air resistance as proportional to the square of the velocity, the falling trajectory of the UAV is solved in combination with the initial conditions, and the landing time and landing speed of the UAV are obtained. The landing speed includes the horizontal velocity and vertical velocity at the time of falling. S13, calculates the instantaneous kinetic energy of the drone during its fall based on horizontal and vertical velocities. ; S14, The crash zone was determined using the Monte Carlo method, based on the 3... Criteria for classifying crash sites, including , and The fall area was obtained by taking the mean of the longitudinal and lateral distribution of the fall points as the center and taking 1, 2, and 3 times the length of the longitudinal and lateral standard deviations as the major and minor axes, respectively.

[0013] Optionally, S2 includes: S21 divides the process of a drone crash injuring a pedestrian into stages, including the crash, impact with a person, and death, and calculates the risk value for the pedestrian. ; S22 categorizes the damage to roads and vehicles caused by drone crashes into different stages, including system malfunction, collision with vehicles, and damage, and calculates the risk value for roads and vehicles. ; S23 defines pedestrian risk and road vehicle risk as the first type of risk. The second type of risk The first and second risks are normalized, and the total risk cost of the grid cell is calculated. ; S24. Spatially combine the total risk costs of all grid cells to construct a risk matrix. ; S25, sum the total risk costs of all grid cells traversed along the path to obtain the risk value of the path. .

[0014] Optionally, S3 includes: S31, by transforming the risk distribution map into a high-density grid map model, the space is discretized, with each grid cell representing a spatial location, recording whether the drone passes through that location, forming a path matrix. Simultaneously, a matrix of risk values ​​for each grid cell by the drone is constructed. The total risk value of the flight path is calculated using matrix multiplication; S32 introduces flight path constraints; S33, dynamically adjust the path expansion step size according to the density of obstacles around the current node; S34, a hybrid heuristic function is constructed by weighted combination of Euclidean distance and Manhattan distance; S35: After the path is generated, the path is simplified and redundant intermediate nodes are removed through feasibility verification. S36, Calculate the number of sampling points based on the maximum coordinate interval of the three spatial axes. .

[0015] Optionally, the flight path constraints in S32 include: Obstacle avoidance constraint: Requires that no point in the path overlaps with a known obstacle area; Flight range constraints: These constraints require that waypoints must be within the defined three-dimensional flight boundary. Flight duration constraint: Limits the number of waypoints along the entire route to a given upper limit; Task termination constraint: The endpoint of the path must be exactly the same as the task objective point.

[0016] Optionally, S4 includes: S41. During the flight of the UAV, if the risk value of the waypoint is not lower than the dynamic risk threshold due to environmental changes, the local path adjustment mechanism is activated. The path is replanned based on the A* algorithm with dynamic weight and adaptive step size to avoid high-risk areas. If the risk value of the adjusted path is still not lower than the dynamic risk threshold, the UAV hovers and waits until the path meets the requirements before continuing to fly. S42, in the local path adjustment, adopts a fast expansion random tree algorithm based on risk perception and incremental adjustment. By introducing a risk perception cost function and a dynamic step size control mechanism, the step size is adjusted according to the changing trend of the risk value at the current position when each node is expanded. If the risk value in the expanded path exceeds the static threshold, the expansion is automatically terminated.

[0017] Optionally, the fast expanding random tree algorithm based on risk perception and incremental adjustment includes: The cost function of the path is extended from distance optimization to a weighted combination that considers both distance and risk value. A dynamic adjustment mechanism for the extended step size based on regional risk factors is introduced to adaptively adjust the step size according to the risk value of the region where the current node is located. The fast expanding random tree algorithm based on risk perception and incremental adjustment has a path replanning strategy, including path risk detection, risk exceeding the threshold triggering replanning, and multiple replanning mechanisms.

[0018] The UAV ground-to-ground risk mitigation device, used to implement the aforementioned UAV ground-to-ground risk mitigation method, includes the following modules: The first processing module acquires the fault status of the UAV during flight. Based on the fault status, it uses a UAV crash model based on ballistic descent to predict the crash area and the probability of impact in different areas after the UAV malfunctions. The second processing module quantifies the risk value of the area based on the relevant factors affecting the degree of risk in the area, and weights and integrates the impact probability of different fall areas with the risk value in the area to generate a risk distribution map. The third processing module: Based on the quantitative assessment results of UAV ground risk in the risk distribution map, the A* algorithm based on dynamic weights and adaptive step size is used to plan the global flight path of the UAV. The fourth processing module: During the operation of the UAV along the initial global flight path, the risk value in the dynamic environment is evaluated in real time based on the risk distribution map. When the risk value at the path point is detected to be no less than the dynamic risk threshold, the online path local optimization mechanism is activated. The risk perception cost function and dynamic step size control are constructed through the fast extended random tree algorithm based on risk perception and incremental adjustment. The flight path from the current point to the target location is recalculated and the UAV flies accordingly.

[0019] The UAV ground-to-ground risk mitigation system includes a memory and a processor. The memory stores a computer program that is executed by the processor. When the computer program is run by the processor, it executes the aforementioned UAV ground-to-ground risk mitigation method.

[0020] A storage medium for mitigating ground-to-air risks during drone operation is provided, wherein a computer program is stored on the storage medium, and the computer program executes the aforementioned method for mitigating ground-to-air risks during drone operation when it is run.

[0021] The beneficial effects of this invention are: This invention predicts the crash area and its impact probability by using a UAV crash model based on ballistic descent. It combines risk quantification analysis of factors such as pedestrians and road vehicles on the ground, and generates a risk distribution map by weighting and fusing the impact probability with the regional risk value. Then, it uses the A* algorithm based on dynamic weights and adaptive step size for global path planning, which can effectively reduce the ground risk during UAV operation, improve the global safety of the flight path, and avoid potential damage to ground personnel, vehicles and facilities.

[0022] This invention dynamically assesses path point risks based on a risk distribution map during flight. When the risk value exceeds a set threshold, a local path optimization mechanism is triggered. A risk perception cost function and dynamic step size control are introduced using a fast expanding random tree algorithm based on risk perception and incremental adjustment to achieve rapid reconstruction and optimization of local paths. This effectively addresses sudden high-risk areas in dynamic environments and ensures the continuous safe operation of UAVs in complex environments. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of the mitigation method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the impact point of the drone when it crashes to the ground, according to an embodiment of the present invention. Figure 3 This is a schematic diagram of the area distribution of the UAV impact area under different probability density functions according to an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the safety risks of drones to pedestrians on the ground according to an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the safety risks of drones to road vehicles according to an embodiment of the present invention; Figure 6 This is a schematic diagram of risk distribution according to an embodiment of the present invention; Figure 7 This is a schematic diagram of the improved A* algorithm based on dynamic weights and adaptive step size according to an embodiment of the present invention; Figure 8 This is a schematic diagram of the real-time adjustment process of the local operating path of a UAV based on dynamic risk perception, according to an embodiment of the present invention. Figure 9 This is a schematic diagram of the functional modules of the device according to an embodiment of the present invention. Detailed Implementation

[0025] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. Those skilled in the art may employ other alternative methods to implement some well-known technologies; moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0026] like Figures 1-8As shown, the method for mitigating ground-to-ground risks from unmanned aerial vehicle (UAV) operations includes the following steps: S1. Obtain the fault status during the flight of the UAV; based on the fault status, predict the possible crash area distribution of the UAV using a UAV crash model based on ballistic descent, and generate a crash area distribution map and corresponding probability values. S2 delves into the key factors influencing ground risks, elucidating the impact mechanisms and coupling mechanisms between these key factors and ground risks. Based on relevant factors affecting risks within areas where drones may crash, a risk assessment model is constructed to quantify the ground risk value of that area. Combining the distribution map and probability values ​​of drone crash areas, the probability values ​​of crash areas and the ground risk values ​​of the areas are weighted and integrated to generate a risk distribution map. S3. To reduce the ground-to-ground risk during UAV operation, a global flight path planning model for UAV is constructed based on the quantitative assessment results of UAV ground-to-ground risk in the risk distribution map. The A* algorithm based on dynamic weights and adaptive step size is used to solve the global flight path planning model for UAV to obtain the initial global flight path. In S4, as the UAV flies along the initial global path, dynamic environmental changes alter the ground risk value of that path. If these changes increase the ground risk value in a local area of ​​the path, the risk of the current path is assessed in real time based on the risk distribution map constructed in S2. When the risk value of a certain segment exceeds a preset threshold, a local path dynamic adjustment is immediately initiated within the initial global path framework, starting from the current location and ending at the original target point. With the core objective of avoiding high-risk areas, a local real-time replanning mechanism based on a fast expanding random tree algorithm grounded in risk perception and incremental adjustment is designed to achieve real-time perception of sudden risks and precise adjustment of local flight paths.

[0027] S1 includes: S11, obtain the fault status of the drone during flight; S12, Construct a drone crash model based on ballistic descent.

[0028] S11. Based on the positioning error, UAV system reliability, operator factors, meteorological factors, airspace type and flight mode during UAV flight, the fault status during UAV flight is obtained.

[0029] Positioning errors are a common problem in drone flight, especially when navigation relies on GPS or other positioning systems. Positioning errors can cause flight path deviations, increasing the risk of collisions with the ground or obstacles. Positioning errors can be categorized into systematic errors and random errors. Errors caused by factors such as weather interference and misinterpretation are typically systematic errors, whose effects are persistent and severe, while errors caused by factors such as design flaws are random errors. Compared to random errors, systematic errors usually have a larger error range and are more dangerous, thus requiring in-depth analysis. Currently, systematic errors are caused by the following factors: Satellite ephemeris error: There is a discrepancy between the actual position of the satellite and the position calculated by ground monitoring equipment. Satellite ephemeris is a key parameter for calculating satellite position, and errors can affect the accuracy of UAV positioning.

[0030] Satellite clock bias: The difference between satellite atomic clocks and GPS standard time can cause errors in transmission time, which can affect the distance calculation between satellites and drones, and thus affect the accuracy of positioning.

[0031] Satellite signal propagation error: Satellite signals need to pass through the ionosphere and troposphere during transmission. The characteristics of the ionosphere and troposphere affect the propagation speed of electromagnetic waves, leading to signal delay and positioning errors.

[0032] Geometric position error: The satellite signals received by the UAV contain geometric position errors. Because different satellites have different geometric positions, the magnitude of the positioning error will vary, affecting the final navigation accuracy.

[0033] Meteorological factors are a crucial external condition affecting drone flight, especially under extreme weather conditions. Weather changes can significantly impact the flight stability and safety of drones, and may even lead to crashes. The main impacts of meteorological factors on drone flight are as follows: Wind speed and direction: Strong winds and gusts are common meteorological factors that cause drones to lose control and crash. In high wind conditions, especially crosswinds or vertical winds, the flight attitude of the drone may change, increasing the difficulty of control. Especially when the drone is small or has poor wind resistance, strong winds may prevent the drone from maintaining the planned flight path, or even cause the aircraft to stall or turn around, resulting in a crash.

[0034] Precipitation: Rain, snow, and other precipitation phenomena increase the weight of drones, leading to increased flight load, especially during long-haul flights. Moisture and water droplets may penetrate electrical components, causing short circuits or damage. At the same time, precipitation reduces the working accuracy of drone sensors, affecting the accuracy of positioning systems and vision sensors.

[0035] Fog and low visibility: In dense fog or low visibility conditions, the effectiveness of a drone's visual sensors and navigation systems is greatly reduced. The drone's obstacle avoidance system may fail to accurately identify obstacles or ground features, thus increasing the risk of collision.

[0036] Temperature and air pressure changes: Extreme temperature changes can affect drone battery performance, especially in low temperatures. Battery life decreases, leading to insufficient power supply or even battery failure, potentially causing the drone to crash. Changes in air pressure can also affect the drone's barometric pressure sensors, causing inaccurate altitude readings and potentially resulting in flying too low or too high, increasing the risk of accidents.

[0037] The reliability of an unmanned aerial vehicle (UAV) system refers to its ability to maintain stable operation of its key components and systems during flight. Insufficient system reliability may lead to malfunctions during mission execution, increasing the risk of crashes. Common types of malfunctions include power system problems (such as insufficient battery power or motor failure), flight control system failures (such as sensor malfunctions or signal loss), hardware aging and improper maintenance, and software errors. These malfunctions can cause unstable flight, even loss of control or deviation from the flight path, ultimately resulting in a crash. To ensure safe flight of UAVs, it is necessary to strengthen the overall reliability design of the system, conduct regular inspections, and perform timely maintenance to reduce the probability of malfunctions and ensure flight safety.

[0038] Operator factors involve the drone operator's skills, experience, and judgment. During flight, operator errors, fatigue, inattention, or lack of familiarity with the equipment can cause the drone to go out of control. Common mistakes include insufficient pre-flight equipment checks, misoperation of the flight control system, failure to respond promptly to emergencies, or misjudgment of the flight environment. Especially in complex flight missions, the operator's decision-making and reaction speed are crucial to flight safety. If the operator fails to effectively respond to environmental changes or sudden malfunctions, the drone may deviate from its intended flight path and ultimately crash.

[0039] Airspace type and flight mode are crucial factors affecting drone flight safety, with different airspace characteristics and flight modes posing unique safety risks. Drones typically operate in low-altitude airspace, generally referring to areas below 1000 meters in altitude, while ultra-low-altitude airspace often refers to areas below 120 meters, an altitude usually corresponding to the visual line-of-sight (VLOS) flight limit. Drone flight modes include VLOS and beyond-visual-line-of-sight (BVLOS), and their impact on safety cannot be ignored. In VLOS mode, the pilot can directly visually observe the drone, monitor its status in real time, and make operational adjustments, effectively reducing the likelihood of accidents. In contrast, in BVLOS mode, the pilot cannot directly observe the drone and must rely on remote sensing equipment or navigation systems for control. This not only increases the difficulty of responding to emergencies but may also lead to signal delays or equipment malfunctions causing the drone to go out of control or even fail to avoid obstacles in time. In addition, the presence of complex elements such as power lines, trees, and buildings in low-altitude airspace further exacerbates the challenges of BVLOS flight.

[0040] S12 includes: By employing the UAV ballistic descent mode and combining it with modeling analysis, the approximate range of the UAV's ground impact point can be determined, thus accurately studying the UAV's crash situation after experiencing an in-flight malfunction.

[0041] The crash patterns associated with internal malfunctions of unmanned aerial vehicle (UAV) systems include free fall or projectile motion in a state of loss of control or power. Considering air resistance during the UAV's descent, the crash process is modeled. The motion of the UAV after crashing is decomposed into horizontal and vertical directions, resulting in the following equations of motion: (1) in, For drone quality; The force of gravity acting on the drone; The horizontal displacement of the drone during its crash. The vertical displacement of the drone during its fall; The time of the drone's crash; The air resistance experienced by the drone in the horizontal direction; The air resistance experienced by the drone in the vertical direction.

[0042] An analysis of the air resistance experienced by the drone reveals the following: the horizontal and vertical resistances are as follows: (2) in, This refers to the air drag coefficient; air density; The frontal area of ​​the drone from the side; This refers to the windward area of ​​the drone in the vertical direction.

[0043] Substituting equation (2) into equation (1), we get: (3) Where g is the acceleration due to gravity.

[0044] Based on the initial conditions when the drone malfunctioned and crashed, i.e. and ,in, This indicates the initial velocity at the point of failure when the drone malfunctions. This indicates the initial lateral velocity at the point of failure when the drone malfunctions. This represents the initial velocity in the vertical direction at the point of failure when the drone malfunctions. This indicates the lateral location of the point of failure when the drone malfunctions. This represents the vertical position of the point of failure when the UAV malfunctions. Integrating equation (3) and substituting the initial conditions, we obtain the location of the UAV crash point as follows: (4) The drone's operating altitude is known to be At that time, its fall and landing time is as follows: (5) At this moment, the instantaneous velocity of the drone upon impact is: (6) in, The horizontal velocity of the drone at the time of crash. This represents the vertical velocity of the drone upon impact.

[0045] From the instantaneous velocity of the drone at the moment of impact, the kinetic energy of the drone at the moment of impact can be obtained as follows: (7) The actual impact points of a crashed drone are randomly distributed around the standard impact point, and their probability density function depends on the statistical distribution of the impact error. Therefore, the Monte Carlo method can be used to determine the area of ​​influence of the ground impact zone. When the drone has an initial horizontal velocity... and height A diagram illustrating the point of impact when an object falls to the ground during flight is shown below. Figure 2 As shown.

[0046] The distribution characteristics of UAV impact points and the area of ​​the affected region show significant differences under different probability density functions. To accurately assess the differences in the probability of UAV impact on different areas of the ground, based on the impact area... The criteria divide the distribution of crash sites into three categories, namely: , and To more comprehensively consider different risk levels, the mean of the longitudinal and lateral distributions of the impact points are used as the center, and the major and minor axes are plotted as 1, 2, and 3 times the length of the longitudinal and lateral standard deviations, respectively, to obtain the different impact areas of the impact point distribution, such as... Figure 4 As shown.

[0047] S2, based on the fault status, a ground pedestrian safety risk assessment and ground risk quantification assessment of the UAV are conducted using a ballistic descent-based UAV crash model, resulting in a risk distribution map. The risk level is quantified based on relevant factors affecting the risk level within the area, and the impact probability of the crash area is weighted and fused with the risk value within the area to generate the risk distribution map.

[0048] S2 uses a ballistic descent-based UAV crash model to predict the potential crash area distribution of UAVs. It quantifies the risk levels of relevant factors within each area, including pedestrian safety risk assessment and vehicle safety risk assessment. Finally, it weights and fuses the risk probabilities with the target risk values ​​to generate a risk distribution map. S21, Ground pedestrian safety risk assessment; The safety risks to pedestrians during drone flights are a crucial and undeniable core issue. With the widespread application of drones in logistics, urban monitoring, and other fields, their flight activities are becoming increasingly frequent, especially in urban environments where flight paths inevitably traverse densely populated areas. Because drones typically operate at lower altitudes, their physical distance to ground targets is significantly shorter compared to traditional aircraft. This means that in the event of mechanical failure or external interference (such as strong winds), drones may quickly become uncontrollable and directly threaten pedestrians. This risk primarily manifests as a drone crashing or colliding with pedestrians when its flight path deviates from its intended trajectory, potentially leading to serious consequences ranging from minor injuries to fatalities. Therefore, the airspace above pedestrians is considered high-risk airspace for drones; if a drone is operating at an altitude of [missing information] above pedestrians, [missing information]. During flight, what are the safety risks to pedestrians on the ground? Figure 6 As shown.

[0049] An incident in which a drone crashes into the ground and causes injury or death requires a series of events to occur. Specifically, it can be broken down into the following three key stages: (1) the drone system fails or goes out of control during flight; (2) the out-of-control drone crashes into the ground and collides directly with at least one pedestrian; (3) the person struck by the drone dies as a result of the impact. Therefore, the probability of injury or death in a drone crash can be expressed as: (8) In the formula: Representative at The risk value of drones to pedestrians on the ground in the area, among which This falls within the area affected by the drone crash. This represents the probability of injury or death to pedestrians when a drone crashes into the ground. This represents the probability of a drone colliding with a pedestrian on the ground.

[0050] The probability of a drone colliding with a pedestrian depends on the area affected by the drone's crash and the population density within that area. (9) In the formula: represent Population density of the area; This represents the probability that a drone will crash in that area. This represents the area affected by the drone crash.

[0051] The probability of injury or death to pedestrians from a drone crash is closely related to the shielding effect of other obstacles such as trees and buildings on the ground, as well as the kinetic energy of the drone upon impact. This can be expressed as: (10) In the formula: This represents the probability of injury or death to pedestrians when a drone crashes into the ground. This falls within the area affected by the drone crash. Represents the impact of a drone crash The impact kinetic energy at that moment; The drone crashed The shading coefficient in the region. represent The impact energy required to achieve a 50% mortality rate at a value of 0.5. Representative The impact energy threshold that causes death to ground personnel when it approaches 0.

[0052] S22, road vehicle safety risks; The potential for uncontrolled crashes of drones flying over roads poses a significant threat to road traffic safety, especially in areas with high pedestrian and vehicular traffic. With the rapid development of drone technology and its widespread application in civilian sectors, such as urban monitoring and emergency rescue, drones are increasingly frequently flying over urban roads, transportation hubs, and highways. This high frequency of operation significantly increases the likelihood of potential conflicts between drones and ground transportation infrastructure, especially in complex urban environments where the density and dynamism of road networks further amplify the risk exposure. The airspace above road lanes is therefore considered a high-risk airspace for drone operations, and its safety issues also involve interactions with the ground transportation system. Therefore, if a drone flies at an altitude of [missing information - likely a specific altitude or height] above road vehicles... The safety risks to vehicles on the road from flying in the air are as follows: Figure 5 As shown.

[0053] Traffic safety risks caused by unmanned aerial vehicles (UAVs) crashing due to loss of control during flight over roads are a critical issue that urgently needs attention in urban airspace management. Similar to the impact of UAVs on pedestrian safety, the safety threat of UAVs to road vehicles can also be described by analyzing three key conditional events: (1) the UAV experiences system failure or loss of control during flight, causing it to deviate from its intended flight path; (2) the UAV crashes to the ground and collides with at least one vehicle; (3) the vehicle that was hit is damaged due to the impact of the UAV, resulting in personal injury or death. Based on these conditional events, the probability of vehicle damage in a UAV failure-to-ground collision accident can be calculated using conditional probability and joint probability: (11) In the formula: Representative at Risk level of drones to road vehicles in the area; This represents the probability that a drone will crash and collide with a vehicle after the drone system fails or goes out of control. The conditional probability of damage to a vehicle in the event of a collision.

[0054] Among them, the probability of a drone crashing and hitting a vehicle on the road. Determined by the ratio of the area covered by vehicles on the road to the total road area: (12) In the formula: This represents the average projected area of ​​a vehicle on the road. Represents road area; This represents the number of vehicles on the road.

[0055] Among them, the number of vehicles on the road Determined by the product of traffic density and road length: (13) In the formula: This represents the vehicle density on the road, that is, the number of vehicles per unit length of road. Represents the length of the road.

[0056] Among them, road area Determined by the product of the road's length and width: (14) In the formula: This represents the width of the road.

[0057] S23, based on the relevant factors affecting the degree of risk within the region, quantifies the degree of risk, and finally weights and merges the risk probability with the target risk value to generate a risk distribution map. For example... Figure 3 As shown.

[0058] In order for the risk distribution map to simultaneously include multiple different types of risk information, it is necessary to adjust and integrate the different risk representation methods. First, let's denote the safety threats posed by drones to pedestrians and road vehicles as risk type 1 and risk type 2, respectively.

[0059] (15) (16) (17) In the formula: This indicates the first risk: the safety threat posed by drones to pedestrians on the ground. The second type of risk is the safety threat posed by drones to road vehicles. The values ​​of the other parameters are referenced from equations (8) to (12).

[0060] In the framework of UAV-based ground risk assessment, normalization serves as a crucial data preprocessing technique. Its core advantage lies in effectively eliminating biases caused by differences in dimensions or magnitudes among various risk indicators by mapping them to a unified numerical range, thereby ensuring the fairness and comparability of the risk assessment. For the comprehensive assessment of personnel risk and road traffic risk, normalization enables the weighted fusion of various risk components under a standardized scale, preventing any single indicator from dominating the overall risk result due to excessive magnitude. This improves the balance of the assessment system and provides a standardized input foundation for subsequent mathematical modeling, risk visualization, and decision support.

[0061] (18) in, These are the risk costs to pedestrians and vehicles on the ground after normalization; These are the calculated risk costs for pedestrians and vehicles on the ground; and These represent the minimum and maximum values ​​of the two risk costs, respectively.

[0062] To balance the weight of different risk objectives in the overall risk cost and ensure the scientific rigor and consistency of the comprehensive risk assessment, a weighting coefficient is introduced. When a grid area in a risk distribution map contains multiple risk sources, the total risk cost of that grid area is: (19) in, The total risk cost of a grid cell; where , If a certain risk source does not exist within the grid area, it is not included in the total risk cost of the area. or .

[0063] To comprehensively characterize the risk characteristics along the UAV's flight path, a comprehensive risk cost model incorporating various risk sources is constructed. A ground risk distribution is built, where each discrete unit has a corresponding risk value. Risk matrix: (20) in, Represents the first in the discrete grid of the ground The overall risk value of each unit.

[0064] Therefore, the total risk cost of a path for: (twenty one) in, This is the set of grid regions traversed by the path.

[0065] Risk distribution map as follows Figure 6 As shown.

[0066] S3 includes: S31. Based on the obtained risk distribution map, construct a discrete path planning model based on mitigation of land risks; S32 uses the A* algorithm based on dynamic weights and adaptive step size to solve the discrete path planning model, and obtains a global flight path with less ground risk for UAV operation.

[0067] S31 includes: Unmanned aerial vehicles (UAVs) are widely used in inspection, logistics, and other fields. These tasks all require flight path planning for UAVs, and a safe and efficient flight path is a prerequisite for the successful completion of these tasks. Discrete UAV global path planning models simplify the path planning problem by dividing a continuous space into a series of discrete grid cells. In this model, the space is divided into a series of fixed-size cells, each representing a specific spatial location. This discretization method reduces the complexity of path planning, allowing the computation to be completed on a finite number of nodes, thus significantly reducing the computational burden. In discrete path planning, the UAV's flight path is achieved through connections between these grid cells. Each grid cell corresponds to a flight state, and the goal of path planning is to find the optimal or shortest path from the starting point to the target point among these states. In the planning process, in addition to considering spatial constraints, it is also necessary to follow rules and constraints during flight, such as avoiding obstacles. This method effectively transforms the complex path planning problem into a search within a discrete grid, thereby improving computational efficiency and the operability of path planning.

[0068] Discrete UAV path planning models include: objective function and constraints.

[0069] (a) Objective Function The objective function is used to evaluate the quality of a path and serves as a standard for measuring path quality in optimization algorithms. The drone's operating environment is divided into two categories: ① Space occupied by obstacles is an insurmountable area, where the drone must maintain a safe distance and avoid collisions by detouring; ② Space without obstacles is a traversable area for the drone, but operational risks must be considered when flying in this area. To minimize operational safety risks while avoiding collisions with obstacles, the objective function is defined as a risk-related expression.

[0070] The information about the drone's flight path can be represented as follows: (twenty two) in, .

[0071] Quantify the risk value of the drone flight area: (twenty three) in, This indicates the ground risk value of the drone in each unit.

[0072] In summary, the objective function established from the perspective of security risk is to minimize the total risk cost: (twenty four) in, The total risk cost of the UAV's flight path is the sum of the risk values ​​of each area the UAV passes through. The lower the total risk cost, the lower the risk level to the main ground targets.

[0073] (ii) Constraints Constraints are key limiting factors in the flight of unmanned aerial vehicles (UAVs). They ensure that path planning not only conforms to flight rules but also meets certain specific requirements. These requirements typically include flight altitude, speed limits, obstacle avoidance requirements, geographical restrictions on the flight area, and other safety regulations. In practical applications, constraints can be dynamically adjusted according to mission requirements and environmental changes to ensure the safe flight of UAVs in complex environments.

[0074] (1) Obstacle avoidance constraints The drone must avoid obstacle areas. The obstacle area is defined as... Then each path point Must meet: (25) (2) Unmanned Aerial Vehicle (UAV) Flight Range Constraints Every waypoint the drone flies through must be within a designated airspace, that is: (26) in, , , , , , These are the boundary ranges of the drone's flight area; This represents the side length of the cell.

[0075] (3) Unmanned aerial vehicle (UAV) flight time constraints Due to the limited battery life of drones, flight time needs to be constrained. This constraint can be achieved by limiting the number of waypoints the drone passes through. Specifically, the total number of waypoints traversed by any complete flight path must not exceed a set maximum. This ensures that the drone can complete its mission within the battery's range, avoiding mission failure or safety incidents due to insufficient power. At the same time, reasonable waypoint restrictions help optimize flight routes, reduce unnecessary detours, and thus improve mission efficiency.

[0076] (27) in, The maximum number of waypoints allowed for a single path; The total number of waypoints in a path; (4) Task termination constraints (28) in, The coordinates of the last waypoint in the drone's flight path. The coordinates are the endpoint coordinates.

[0077] In S32, global path planning in UAV path planning refers to pre-calculating the UAV's flight path based on the starting and destination locations before the flight mission begins, ensuring that it can avoid obstacles, meet flight rules, and reach the destination. The A* algorithm based on dynamic weights and adaptive step size is used to solve the discrete UAV global path planning model, obtaining paths with lower ground-to-ground risks for UAV operation. The A* algorithm based on dynamic weights and adaptive step size introduces three key improvements over the traditional A* algorithm to enhance the search efficiency, path quality, and applicability of path planning, specifically: (1) Constructing a dynamic weight adjustment mechanism The fixed weights of the traditional A* algorithm cannot adapt to the needs of different environments. The A* algorithm based on dynamic weights and adaptive step sizes dynamically adjusts heuristic weights, enabling the search strategy to adapt to environmental complexity. In open areas, weight values ​​are reduced to improve search efficiency; in complex environments, weight values ​​are increased to refine the search and improve path quality.

[0078] The objective function of the A* algorithm based on dynamic weights and adaptive step size was modified by introducing dynamic weight coefficients. : (29) Among them, dynamic weight coefficient The calculation formula is dynamically adjusted according to the search process: (30) In the formula: and These are the upper and lower limits of the dynamic weighting coefficients, respectively. The current node To the starting point Euclidean distance; It is the starting point To the finish line Euclidean distance.

[0079] This dynamic weighting mechanism enables the algorithm to perform well in the early stages of the search, especially when near the starting point. near The algorithm tends to move quickly towards the target, but in the later stages of the search, near the endpoint, near The algorithm tends to find more precise paths. In this study, we set... , This allows the algorithm to automatically adjust its search strategy during the search process, achieving a better balance.

[0080] (2) Adaptive step size adjustment The A* algorithm, based on dynamic weights and adaptive step size, introduces an adaptive step size strategy, dynamically adjusting the step size according to the openness of the current environment. In open areas, the step size is larger to reduce redundant calculations and improve search speed; in areas with dense obstacles, the step size is smaller to improve the accuracy of path planning and ensure that the UAV can safely avoid obstacles.

[0081] To address the limitations of a fixed step size, the A* algorithm, based on dynamic weights and an adaptive step size, introduces an adaptive step size mechanism. The step size calculation formula is as follows: (31) In the formula: Represents the current step size. It is a node Surrounding radius The number of obstacles inside; It is the detection radius; It is the attenuation coefficient; It is the maximum step size.

[0082] (3) Hybrid heuristic function design The A* algorithm based on dynamic weights and adaptive step size uses a hybrid of Euclidean distance and Manhattan distance as its heuristic function: (32) In the formula: It is Euclidean distance; It's Manhattan distance; These are the mixing coefficients, which are related to the dynamic weights: (33) (4) Path optimization and redundant point removal The A* algorithm, based on dynamic weights and adaptive step sizes, uses a path optimization strategy to reduce the number of path points after generating the initial path. The optimization idea is to find the farthest node that can be directly reached from the current node and skip intermediate points.

[0083] Let the initial path be: (34) The optimized path is: (35) In the formula: satisfy: (36) Where: Path validity check function for: (37) In the formula: for For each sampling point in the set, if from arrive None of the points sampled between are obstacles. If the path is valid, then the path is valid; otherwise, the path is invalid.

[0084] Number of sampling points for: (38) In the formula: This represents the absolute value of the distance between two points along the x-axis. This represents the absolute value of the distance between two points along the y-axis. This represents the absolute value of the distance between two points along the z-axis.

[0085] The flowchart of the A* algorithm based on dynamic weights and adaptive step size is as follows: Figure 7 As shown. Its process is described below; (1) Initialization: Create an open list to store nodes to be expanded and a closed list to store expanded nodes. Define the starting and target nodes for the search. Set the initial weight factors. This is used to adjust the degree of influence of the heuristic function.

[0086] (2) Calculate the node evaluation function.

[0087] Set the starting node A value of 0 indicates that the cost from the starting point to the target point is 0; the heuristic estimate of the cost from the starting node to the target point is calculated according to equation (32). And calculate the total estimated cost of the node according to equation (29). Dynamic weighting coefficients The algorithm is dynamically adjusted as the search progresses and updated according to equation (30).

[0088] (3) Loop execution: Select values ​​from the open list The smallest node is selected as the current node. If the current node is the target node, a path is constructed and returned, and the current node is removed from the open list and added to the closed list. Otherwise, the neighboring nodes are expanded by an adaptive step size (Equation 31).

[0089] (4) Processing neighboring nodes: For each neighboring node of the current node, skip it if the neighboring node is already in the closed list. Calculate the distance from the current node to the neighboring node. If a neighboring node is not in the open list, add it to the open list and record its parent node. If a neighboring node is already in the open list, check if the path from the current node to the neighboring node is shorter. If it is shorter, update its value. Value and parent node.

[0090] (5) Path construction: When the target node is found, backtrack from the target node to the starting point to obtain the initial path. After generating the initial path, use the path optimization strategy to reduce the number of path points.

[0091] S4, a method for real-time adjustment of the local operating path of UAVs considering dynamic risk perception.

[0092] During the operation of a UAV along its initial global flight path, changes in the dynamic environment can alter the global path's ground-to-ground risk value. Based on the risk map in S2, the dynamic risk value of the environment is assessed, and a real-time local path adjustment mechanism for the UAV is designed. When the ground-to-ground risk value at a path point exceeds a set threshold, an online local path optimization mechanism is initiated. A fast expanding random tree algorithm based on risk perception and incremental adjustment is proposed for online local optimization. With the goal of minimizing the ground-to-ground risk value, a risk perception cost function and dynamic step size control are constructed. The flight path from the current point to the target location is recalculated and followed, achieving real-time perception of sudden risks and precise adjustment of the local flight path.

[0093] S41, Dynamic Risk Perception and Real-time Adjustment Mechanism for Local Drone Paths; Based on global path planning, the UAV flies along a predetermined path. During flight, dynamic environmental changes such as sudden obstacles or flight restrictions may alter its ground risk value, necessitating adjustments to the global path based on real-time ground risk values. Online path planning needs to respond quickly to environmental changes and generate new feasible paths promptly. Building upon the A* algorithm planning based on dynamic weights and adaptive step sizes, a risk-aware-based real-time adjustment mechanism for the UAV's local path is proposed, aiming to reduce the risk threat to key ground targets throughout the flight. The adjustment mechanism process is described below. Figure 8 The specific description is as follows: An A* algorithm based on dynamic weights and adaptive step size is used to generate the initial flight path, which consists of multiple ordered waypoints. The UAV passes through each waypoint sequentially. Upon reaching a waypoint, the risk value is detected using a UAV-to-ground risk assessment model. A dynamic risk threshold Rt is defined based on urban low-altitude safety standards. When the UAV reaches a waypoint, the risk value R of the current waypoint is calculated based on the ground risk assessment model. When R ≥ Rt, a local real-time path adjustment algorithm is triggered to find the optimal path to avoid high-risk areas from the current waypoint. After real-time path adjustment, if the risk value of the new waypoint still exceeds the threshold, online path replanning is restarted until the UAV safely reaches the target location.

[0094] S42, a fast expanding random tree algorithm based on risk perception and incremental adjustment; To address the limitations of traditional Rapidly-exploring Random Tree (RAR) algorithms in high-dimensional path planning, a risk-aware, incrementally adjusted RAR algorithm is proposed. This algorithm introduces risk-aware path cost calculation, dynamic step size adjustment, and multiple path replanning strategies. Risk-aware path cost calculation is achieved by integrating distance and risk costs and balancing them with weighting factors. The step size is adjusted based on the current node's risk value, with smaller step sizes in high-risk areas and larger step sizes in low-risk areas. Risk values ​​are monitored in real-time, triggering replanning when thresholds are exceeded, until the drone reaches the target location, thereby improving the safety and reliability of path planning.

[0095] The fast expanding random tree algorithm based on risk perception and incremental adjustment mainly includes the following improvements: (1) Risk perception path cost calculation: By conducting risk assessment on the path, the path can avoid passing through high-risk areas, thereby enhancing the path's ground security.

[0096] In the fast expanding random tree algorithm based on risk perception and incremental adjustment, the cost of a path is not only determined by the distance between nodes, but also by the potential risk value of each path segment. Specifically, the path cost function is calculated using the following formula: (39) In the formula: It is the path cost; It is the Euclidean distance of the path segment; It is the risk and cost of the path segment; This is a weighting factor that controls the balance between distance and risk. Using this formula, the fast expanding random tree algorithm based on risk perception and incremental adjustment can effectively consider the risks in different areas of the path. Especially in high-risk areas, the algorithm tends to choose safer paths, thereby avoiding potential collision risks.

[0097] (2) Dynamic Step Size Adjustment: The step size of the expanded tree node is adjusted according to the risk level of the environment to avoid generating excessively large step sizes in high-risk areas. Traditional fast expanded random tree algorithms typically use a fixed step size for node expansion during the tree expansion process, and their design is primarily aimed at static environments, without directly considering the impact of dynamic environmental changes. However, in actual UAV application scenarios, the distribution of obstacles and risk areas in the environment often exhibit dynamic characteristics, which places higher demands on the adaptability and real-time performance of path planning algorithms. Therefore, traditional fast expanded random tree algorithms may exhibit certain limitations when facing dynamic environments, failing to adjust the path in a timely manner to avoid newly emerging obstacles or high-risk areas. To address this issue, the fast expanded random tree algorithm based on risk perception and incremental adjustment introduces a dynamic step size adjustment mechanism. Specifically, when expanding tree nodes, the step size is adjusted according to the risk value of the area where the current node is located: (40) In the formula: It is the original step size; This is the current path point-to-ground risk value. When a node is in a high-risk area, the step size is reduced to prevent excessively large step sizes from leading to more dangerous areas. Conversely, in low-risk areas, the step size remains larger to accelerate path expansion.

[0098] (3) Multiple Path Replanning: A key feature of the Fast Expanding Random Tree Algorithm based on Risk Perception and Incremental Adjustment is its support for multiple path replanning. In dynamic environments, obstacles may change, especially when UAVs are performing missions, they may encounter new high-risk areas. Therefore, the Fast Expanding Random Tree Algorithm based on Risk Perception and Incremental Adjustment implements dynamic path adjustment through the following strategies: Path detection: During the drone's flight along the path, the risk value of the current path segment is detected in real time. If the risk value of a certain path segment exceeds a set threshold, path replanning will be triggered.

[0099] Replanning process: A fast expanding random tree algorithm based on risk perception and incremental adjustment is used to replan the optimal path from the current path segment to the destination. Replanning considers not only distance factors but also path safety, avoiding crossing high-risk areas.

[0100] Multiple replanning mechanism: If the risk value of the current path segment is higher than the set threshold, the algorithm will replan the path multiple times to ensure that the drone can quickly adjust its flight path and avoid entering high-risk areas.

[0101] like Figure 9 As shown, the UAV ground-attack risk mitigation device, used to implement the aforementioned UAV ground-attack risk mitigation method, includes the following modules: The first processing module is used to obtain the fault status of the UAV during flight. Based on the fault status, it uses a UAV crash model based on ballistic descent to predict the crash area and the probability of impact in different areas after the UAV malfunctions. The second processing module is used to quantify the risk value of the area based on the relevant factors affecting the risk level within the area, and to weight and fuse the impact probability of different fall areas with the risk value within the area to generate a risk distribution map. Among them, the quantification of the risk level of the relevant factors affecting the risk level within the area includes: ground pedestrian safety risk value assessment and road vehicle safety risk assessment. The third processing module is used to perform global flight path planning for the UAV based on the quantitative assessment results of the UAV's ground risk and using the A* algorithm based on dynamic weights and adaptive step size. The fourth processing module is used to trigger a real-time local path adjustment mechanism when the UAV flies along a predetermined global flight path. If the path risk value increases due to dynamic environmental changes, the local path adjustment mechanism is activated. This mechanism employs a fast expanding random tree algorithm based on risk perception and incremental adjustment to optimize the local route in real time, achieving real-time perception of sudden risks and accurate correction of the flight path. Specifically, the dynamic risk value of the dynamic environment is assessed based on a risk map, and a real-time local path adjustment mechanism for the UAV is designed. When the ground risk value at a path point exceeds a set threshold, the online local path optimization mechanism is activated. The risk perception cost function and dynamic step size control are constructed through the fast expanding random tree algorithm based on risk perception and incremental adjustment, and the flight path from the current point to the target location is recalculated and followed.

[0102] The UAV ground risk mitigation system includes a memory and a processor. The memory stores a computer program that is executed by the processor. When the computer program is run by the processor, it executes the aforementioned UAV ground risk mitigation method.

[0103] The storage medium for mitigating ground-to-air risks during drone operation stores a computer program that executes the aforementioned method for mitigating ground-to-air risks during drone operation.

[0104] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0105] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for mitigating ground-to-ground risks associated with unmanned aerial vehicle (UAV) operations, characterized in that, Includes the following steps: S1, obtain the fault status of the UAV during flight, and based on the fault status, predict the crash area and the probability of impact in different areas after the UAV malfunctions by using a UAV crash model based on ballistic descent. S2, quantify the risk value of the area based on the relevant factors affecting the risk level in the area, and weight and integrate the impact probability of different fall areas with the risk value in the area to generate a risk distribution map; S3. Based on the quantitative assessment of UAV ground risk in the risk distribution map, the A* algorithm based on dynamic weights and adaptive step size is used to plan the global flight path of the UAV. S4. During the operation of the UAV along the initial global operating path, the risk value in the dynamic environment is evaluated in real time based on the risk distribution map. When the risk value at the path point is detected to be no less than the dynamic risk threshold, the online path local optimization mechanism is activated. The risk perception cost function and dynamic step size control are constructed through the fast extended random tree algorithm based on risk perception and incremental adjustment. The flight path from the current point to the target location is recalculated and the UAV flies accordingly.

2. The method for mitigating ground-to-ground risks during UAV operation according to claim 1, characterized in that, S1 includes: S11 determines the fault status of the UAV by collecting factors that affect its reliability during operation, including positioning error identification, weather disturbance identification, system and equipment reliability assessment, operator factor analysis, and flight airspace and mode determination. The positioning error identification and analysis are caused by navigation errors due to GPS ephemeris errors, clock errors, signal propagation errors, and geometric configuration errors. The meteorological disturbance identification and assessment method evaluates the impact of environmental factors such as wind speed and direction, precipitation, haze, temperature and air pressure changes on the flight stability of the UAV. The system and equipment reliability assessment includes identifying abnormal conditions such as battery power, flight control failure, communication interruption, and aging components. The operator factor analysis considers flight anomalies caused by human factors such as lack of experience, misoperation, and slow response. The determination of flight airspace and mode is based on risk level classification according to whether it is low-altitude, ultra-low-altitude, VLOS or BVLOS mode. S12. Using the UAV ballistic descent mode and combined with modeling analysis, the falling motion of the UAV after a malfunction is analyzed. By establishing the force and motion equations in the horizontal and vertical directions, and expressing air resistance as proportional to the square of the velocity, the falling trajectory of the UAV is solved in combination with the initial conditions, and the landing time and landing speed of the UAV are obtained. The landing speed includes the horizontal velocity and vertical velocity at the time of falling. S13, based on horizontal and vertical velocities, calculates the instantaneous kinetic energy of the drone during its fall. ; S14, The crash zone was determined using the Monte Carlo method, based on the 3... Criteria for classifying crash sites, including , and The fall area was obtained by taking the mean of the longitudinal and lateral distribution of the fall points as the center and taking 1, 2, and 3 times the length of the longitudinal and lateral standard deviations as the major and minor axes, respectively.

3. The method for mitigating ground-to-ground risks during UAV operation according to claim 2, characterized in that, S2 includes: S21 divides the process of a drone crash injuring a pedestrian into stages, including the crash, impact with a person, and death, and calculates the risk value for the pedestrian. ; S22 categorizes the damage to roads and vehicles caused by drone crashes into different stages, including system malfunction, collision with vehicles, and damage, and calculates the risk value for roads and vehicles. ; S23 defines pedestrian risk and road vehicle risk as the first type of risk. The second type of risk The first and second risks are normalized, and the total risk cost of the grid cell is calculated. ; S24. Spatial combination of the total risk cost of all grid cells to construct a risk matrix. ; S25, sum the total risk costs of all grid cells traversed along the path to obtain the risk value of the path. .

4. The method for mitigating ground-to-ground risks during UAV operation according to claim 3, characterized in that, S3 includes: S31, by transforming the risk distribution map into a high-density grid map model, the space is discretized, with each grid cell representing a spatial location, recording whether the drone passes through that location, forming a path matrix. Simultaneously, a matrix of risk values ​​for each grid cell by the drone is constructed. The total risk value of the flight path is calculated using matrix multiplication; S32 introduces flight path constraints; S33, dynamically adjust the path expansion step size according to the density of obstacles around the current node; S34, a hybrid heuristic function is constructed by weighted combination of Euclidean distance and Manhattan distance; S35: After the path is generated, the path is simplified and redundant intermediate nodes are removed through feasibility verification. S36, Calculate the number of sampling points based on the maximum coordinate interval of the three spatial axes. .

5. The method for mitigating ground-to-ground risks during UAV operation according to claim 4, characterized in that, The flight path constraints in S32 include: Obstacle avoidance constraint: Requires that no point in the path overlaps with a known obstacle area; Flight range constraints: These constraints require that waypoints must be within the defined three-dimensional flight boundary. Flight duration constraint: Limits the number of waypoints along the entire route to a given upper limit; Task termination constraint: The endpoint of the path must be exactly the same as the task objective point.

6. The method for mitigating ground-to-ground risks during UAV operation according to claim 5, characterized in that, S4 includes: S41. During the flight of the UAV, if the risk value of the waypoint is not lower than the dynamic risk threshold due to environmental changes, the local path adjustment mechanism is activated. The path is replanned based on the A* algorithm with dynamic weight and adaptive step size to avoid high-risk areas. If the risk value of the adjusted path is still not lower than the dynamic risk threshold, the UAV hovers and waits until the path meets the requirements before continuing to fly. S42, in the local path adjustment, adopts a fast expansion random tree algorithm based on risk perception and incremental adjustment. By introducing a risk perception cost function and a dynamic step size control mechanism, the step size is adjusted according to the changing trend of the risk value at the current position when each node is expanded. If the risk value in the expanded path exceeds the static threshold, the expansion is automatically terminated.

7. The method for mitigating ground-to-ground risks during UAV operation according to claim 6, characterized in that, The fast expanding random tree algorithm based on risk perception and incremental adjustment includes: The cost function of the path is extended from distance optimization to a weighted combination that considers both distance and risk value. A dynamic adjustment mechanism for the extended step size based on regional risk factors is introduced to adaptively adjust the step size according to the risk value of the region where the current node is located. The fast expanding random tree algorithm based on risk perception and incremental adjustment has a path replanning strategy, including path risk detection, risk exceeding the threshold triggering replanning, and multiple replanning mechanisms.

8. A device for mitigating ground-to-air risks during UAV operation, used to implement the method for mitigating ground-to-air risks during UAV operation as described in any one of claims 1-7, characterized in that, Includes the following modules: The first processing module acquires the fault status of the UAV during flight. Based on the fault status, it uses a UAV crash model based on ballistic descent to predict the crash area and the probability of impact in different areas after the UAV malfunctions. The second processing module quantifies the risk value of the area based on the relevant factors affecting the degree of risk in the area, and weights and integrates the impact probability of different fall areas with the risk value in the area to generate a risk distribution map. The third processing module: Based on the quantitative assessment results of UAV ground risk in the risk distribution map, the A* algorithm based on dynamic weights and adaptive step size is used to plan the global flight path of the UAV. The fourth processing module: During the operation of the UAV along the initial global flight path, the risk value in the dynamic environment is evaluated in real time based on the risk distribution map. When the risk value at the path point is detected to be no less than the dynamic risk threshold, the online path local optimization mechanism is activated. The risk perception cost function and dynamic step size control are constructed through the fast extended random tree algorithm based on risk perception and incremental adjustment. The flight path from the current point to the target location is recalculated and the UAV flies accordingly.

9. A ground-attack risk mitigation system for unmanned aerial vehicles (UAVs), characterized in that: It includes a memory and a processor, wherein the memory stores a computer program that is executed by the processor, and the computer program, when executed by the processor, performs the ground risk mitigation method for unmanned aerial vehicle operations as described in any one of claims 1-7.

10. A storage medium for mitigating ground-to-ground risks during UAV operation, characterized in that, The storage medium stores a computer program, which, when running, executes the method for mitigating ground-to-air risks of unmanned aerial vehicle operations as described in any one of claims 1-7.

Images