Unmanned aerial vehicle path navigation planning method based on urban low-altitude airspace risk constraint

Abstract

The invention discloses an unmanned aerial vehicle path navigation planning method based on urban low-altitude airspace risk constraints, and belongs to the technical field of unmanned aerial vehicle path planning. Random sampling and node expansion are carried out by adopting a fast exploration random tree algorithm; performing path local optimization by adopting a beetle antennae algorithm, and generating an expansion node; adopting a collision detection function to carry out obstacle detection between the expansion nodes; constructing a tree, and searching and outputting an optimal path; according to the method, the random expansion characteristic of the rapid exploration random tree and the local optimization capability of the longhorn beetle search are combined, so that a better safe path is rapidly generated in a three-dimensional environment, generation of redundant nodes can be effectively reduced, the path planning efficiency and the convergence speed are improved, and the method is more suitable for rapid path planning of the unmanned aerial vehicle in the urban airspace.

Description

Technical Field

[0001] The present invention belongs to the technical field of UAV path planning, and specifically relates to a UAV path navigation planning method based on the risk constraint of urban low-altitude airspace. Background Art

[0002] As the main vehicle of the low-altitude economy, UAVs are widely used in scenarios such as manned transportation and logistics distribution. The urban airspace scenario endows important significance for the rapid path planning of UAVs. Effective path planning can help UAVs achieve safe and efficient flight in complex urban scenarios, avoid collisions, shorten flight time, and ensure the timely completion of tasks.

[0003] Traditional path planning algorithms (such as Rapidly-Exploring Random Tree, RRT) often have problems such as path redundancy, low node expansion efficiency, poor path smoothness, and slow convergence speed when dealing with high-density obstacles, resulting in extended flight time and increased energy consumption of UAVs, and even affecting flight safety. Summary of the Invention

[0004] The purpose of the present invention is to provide a UAV path navigation planning method based on the risk constraint of urban low-altitude airspace for the above deficiencies in the prior art, so as to solve the problems of path redundancy, low node expansion efficiency, poor path smoothness, slow convergence speed, etc. existing in traditional path planning algorithms.

[0005] To achieve the above purpose, the technical solution adopted by the present invention is:

[0006] A UAV path navigation planning method based on the risk constraint of urban low-altitude airspace, which includes the following steps:

[0007] S1. Set the UAV path navigation planning parameters, including the search tree T, the starting point q of the UAV path start and the end point q goal ;

[0008] S2. Randomly sample a node q rand using the Rapidly-Exploring Random Tree algorithm, and search for the node q rand nearest to the node q near in the search tree T, and then expand to generate the node q new ;

[0009] S3. Take the expanded node q new as the longhorn beetle position, update the longhorn beetle position using the longhorn beetle antenna algorithm, and generate the expanded node q bettle ;

[0010] S4. Connect the expanded node q bettle to the expanded node qnew ;

[0011] If there is no obstacle between the connected paths, the expanded node q bettle is added to the search tree T and connected to the node q near , and then S5 is executed; if there is an obstacle between the connected paths, return to execute S3;

[0012] S5. Calculate the distance D between the expanded node q bettle and the end point q goal . If the distance D is less than the threshold ε, output the optimal path; if the distance D is greater than the threshold ε, iteratively execute S2 to S5 until the distance D is less than the threshold ε.

[0013] Furthermore, S2 specifically includes:

[0014] Randomly sample the node q in the urban airspace environment rand , search for the node q rand nearest to the sampled node q in the search tree T near . If the connection line between the node q near and the sampled node q rand has no obstacle, use the path search step size Δq of the rapidly-exploring random tree algorithm to expand from the node q near in the direction of the sampled node q rand to generate the expanded node q new .

[0015] Furthermore, the said S3 includes the following sub-steps:

[0016] S31. Take the expanded node q new as the longhorn beetle position and randomly generate a longhorn beetle antenna random search unit vector;

[0017] S32. According to the longhorn beetle position and the longhorn beetle antenna random search unit vector, calculate the search positions of the two antennae of the longhorn beetle and the cost function of the corresponding search positions;

[0018] S33. Update the longhorn beetle position according to the cost function, and then generate the expanded node q bettle .

[0019] Furthermore, in the said S31, the longhorn beetle antenna random search unit vector is:

[0020]

[0021] In the formula, is the longhorn beetle antenna random search unit vector; rnd(.) is a random vector, and k is the position dimension.

[0022] Furthermore, the said S32 specifically includes:

[0023] Calculate the search positions of the two antennae of the longhorn beetle:

[0024]

[0025] In the formula, X r and X l are the search positions of the right antenna and the left antenna of the longhorn beetle respectively; d is the sensing distance of the longhorn beetle's antenna; X is the position of the expanded node q new ;

[0026] Calculate the cost function f(X) of the search positions of the two antennae of the longhorn beetle:

[0027]

[0028] In the formula, α 1 , α 2 are the weight coefficients of the distance cost and the risk cost respectively; C i , i = 1, 2,..., k, represents the comprehensive risk cost of the UAV on the RP i-1 RP i flight segment.

[0029] Furthermore, the distance cost and the risk cost are respectively:

[0030]

[0031] Among them,

[0032] In the formula, ||RP i-1 , RP i || is the distance that the UAV moves from the route node RP i-1 to RP i ; v UAV is the cruising speed of the UAV; R(i, j, h) is the equivalent safety level of the UAV at the position of (i, j, h); R norm is the normalized risk value of the path in this area; Med(R(i, j, h)) is the median of all values in the risk matrix of this flight segment.

[0033] Furthermore, the equivalent safety level R(i, j, h) is specifically:

[0034] R(i, j, h) = P(LOC)·R f (i, j, h)

[0035] Among them, R f = R i ⊙M

[0036] R i = D peo ·A cas

[0037] Wherein, P(LOC) is the probability that the UAV fails and crashes during flight in the air, and R f (i, j, h) is the final ground risk value when the UAV crashes within the grid at position (i, j, h) on the ground; R f is the final ground risk value matrix; R i is the initial ground risk matrix; D peo is the ground personnel density in the UAV crash area, and A cas is the casualty area area caused by the UAV crash; M is the ground risk mitigation measure effect matrix, M includes two types, M1 and M2. The M1 type is to reduce the number of people directly exposed to the risk area, and the M2 type is to reduce the impact caused by the UAV hitting the ground.

[0038] Furthermore, in step S33, the longhorn beetle position is updated according to the cost function to x t+1 , and then an extended node q bettle is generated, and the longhorn beetle position x t+1 is updated as:

[0039] x t+1 = x t + δ t * d * sign[f(x r ) - f(x l )]

[0040] δ t+1 = δ t * η

[0041] Wherein, x t is the longhorn beetle position before update; δ t is the search step size of the longhorn beetle antenna algorithm at time t, and δ t+1 is the search step size of the longhorn beetle antenna algorithm at time t + 1; η is the reduction coefficient, d is the sensing distance of the longhorn beetle antenna angle; f(x r ) and f(x l ) are the cost functions corresponding to the search positions of the right antenna and the left antenna of the longhorn beetle respectively.

[0042] The UAV path navigation and planning method based on the risk constraint of the urban low-altitude airspace provided by the present invention has the following beneficial effects:

[0043] 1. The present invention uses an improved RRT method (RRT - BAS) optimized by the Beetle Antennae Search (BAS) algorithm to solve the path planning problem of unmanned aerial vehicles (UAVs) in urban airspace. The Beetle Antennae Search algorithm imitates the antenna sensing mechanism of beetles. Based on the detection results of the left and right antennae, it dynamically adjusts its moving direction and step size to gradually find the optimal path. The BAS algorithm adjusts and relocates the nodes of the path, reduces redundant nodes, and decreases the path turning angle to make the path smoother. At the same time, the Beetle Antennae Search algorithm has high search efficiency and local optimization ability, which is suitable for path planning in complex environments and high - dimensional problems, and can effectively make up for the problem of slow convergence speed caused by randomly sampling nodes in the RRT algorithm.

[0044] 2. The present invention can improve the path planning efficiency of UAVs in complex urban airspace environments. This algorithm combines the random expansion characteristics of the Rapidly - exploring Random Tree (RRT) and the directional optimization ability of the Beetle Antennae Search (BAS) to quickly generate a relatively optimal and safe path in a three - dimensional environment. Through this method, it can effectively avoid obstacles, reduce the public safety risks brought by flight, and ensure the feasibility and safety of the path, especially in urban airspace environments with complex terrain, many obstacles, dense ground personnel, and dynamic risk changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a visualization form of the three - dimensional ground risk matrix of the UAV of the present invention.

[0046] Figure 2 It is an analysis of the falling process of the UAV after failure in the present invention.

[0047] Figure 3 It is a flowchart of the UAV path navigation planning method based on urban low - altitude airspace risk constraints in the present invention.

[0048] Figure 4 It is a path result graph of multiple algorithms during the verification of the present invention.

[0049] Figure 5 It is a convergence speed result graph of multiple algorithms during the verification of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0050] The following describes the specific embodiments of the present invention to facilitate those skilled in the art of the present technology to understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art of the present technology, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions created using the concept of the present invention are within the scope of protection.

[0051] Example 1

[0052] This embodiment provides a method for path navigation planning of unmanned aerial vehicles (UAVs) based on risk constraints in urban low-altitude airspace, using an improved Rapidly-Exploring Random Tree (RRT) method optimized by the Beetle Antennae Search (BAS) algorithm. The BAS algorithm imitates the antenna sensing mechanism of beetles and dynamically adjusts its moving direction and step size based on the detection results of the left and right antennae to gradually find the optimal path. The cost function that combines the distance cost and risk cost of the UAV flight path is used as the objective function of the BAS mechanism, and the excellent local optimization ability of the BAS algorithm is utilized to make up for the defects of the RRT algorithm, such as high comprehensive cost and slow convergence speed caused by random sampling. Refer to Figure 3 , and it specifically includes the following contents:

[0053] Step S1, initialization of the Rapidly-Exploring Random Tree algorithm;

[0054] Set the parameters for UAV path navigation planning, including the initialized search tree T, the starting point q start and the ending point q goal ; The starting point q start and the ending point q goal in this implementation are the basis of the entire path planning, and the path extension of RRT will be carried out based on these two positions.

[0055] Step S2, random sampling and node expansion of the Rapidly-Exploring Random Tree algorithm;

[0056] Specifically, in this embodiment, a node q rand is randomly sampled in the urban airspace environment, and the node q rand nearest to the sampled node q near is searched in the search tree T. If the connection line between the node q near and the sampled node q rand has no obstacles, then using the path search step size Δq of the Rapidly-Exploring Random Tree algorithm, an extension is made from the node q near in the direction of the sampled node q rand to generate an extended node q new .

[0057] Step S3, use the Beetle Antennae Search algorithm for local path optimization, update the beetle position, and generate an extended node q bettle , which specifically includes the following sub-steps:

[0058] Step S31, the beetle position is at the extended node q new , and its position is X:

[0059] X(x 1 ,x 2 ,…,x n )

[0060] In the formula, x i is the centroid position of the longicorn at time i.

[0061] Randomly generate a longhorn beetle and randomly search for a unit vector;

[0062]

[0063] In the formula, is the random search unit vector of the beetle whiskers; rnd(.) is a random vector, and k is the position dimension.

[0064] Step S32, calculating the cost function of the search position of the two antennae of the longhorn beetle and the corresponding search position according to the position of the longhorn beetle and the random search unit vector of the longhorn beetle's tentacles;

[0065] Specifically, in order to simulate the activity of beetle antennae, this embodiment provides the search positions of the right and left antennae of the longhorn beetle respectively to simulate the search behavior;

[0066] The search positions of the two antennae of the longicorn are:

[0067]

[0068] Where, X r and X l The search positions of the right and left antennae of the longicorn respectively; d is the sensing distance of the longicorn whiskers; X is the extended node q new location;

[0069] The longicorn determines the next action based on the route cost function sensed by the left and right tentacles, and calculates the expansion q new The cost function f(X r ) and f(X l ); Specifically, the three-dimensional airspace is divided into several unit cubes, assuming that the optional waypoints RP of the drone k As the vertices of the cube, the path of the drone can be expressed as Path = {RP 1 ,RP 2 ,…,RP k}, the route cost function f(X) can be expressed as the sum of a series of segment costs. Taking into account the path length and flight risk, the cost function f(X) of the search position of the two antennae of the longhorn beetle is:

[0070]

[0071] In the formula, α 1 , α 2 Distance cost Risk Cost The weight coefficient of i, where \(i = 1, 2, \ldots, k\), representing the comprehensive risk cost of the UAV on the RP i-1 RP i flight segment.

[0072] The distance cost and the risk cost are respectively:

[0073]

[0074] To ensure that the distance cost and the risk cost are in the same order of magnitude, the present invention normalizes the risk assessment value \(R(i, j, h)\), and \(R\) norm is the normalized risk value of the path in this area, which is represented by the minimum value of the average and median of all values in the risk matrix of this flight segment, representing the concentrated distribution trend of the risk level in this area:

[0075]

[0076] In the formula, \(\vert\vert RP\) i-1 , RP i \(\vert\vert\) is the distance that the UAV moves from the route node \(RP\) i-1 to \(RP\) i ; \(v\) UAV is the cruising speed of the UAV; \(R(i, j, h)\) is the equivalent safety level of the UAV at the position with coordinates \((i, j, h)\); \(R\) norm is the normalized risk value of the path in this area; \(Med(R(i, j, h))\) is the median of all values in the risk matrix of this flight segment.

[0077] The specific calculation process of the equivalent safety level \(R(i, j, h)\) in this embodiment is as follows:

[0078] The safety risk brought by the operation of the UAV is mainly the expected number of casualties caused by the UAV system failing in the air and crashing on the urban ground (hereinafter referred to as "ground risk" for short), which is the most important impact on public safety when the UAV flies in the urban airspace. In the urban airspace, a three-dimensional safety risk matrix \(R\) n,m,l \((i = 1, 2, \ldots, n; j = 1, 2, \ldots, m; h = 1, 2, \ldots, l)\) can be constructed, as Figure 1 shown:

[0079]

[0080] In the formula, \(R(i, j, h)\) represents the equivalent safety level of the UAV at the position with coordinates \((i, j, h)\), in units of number of casualties per flight hour, and can be calculated by the following formula:

[0081] \(R(i, j, h)=P(LOC)\cdot R\) f(i, j, h)

[0082] Wherein, P(LOC) is the probability that the UAV fails and crashes during flight in the air, and R f (i, j, h) is the final ground risk value when the UAV crashes within the grid at position (i, j, h) on the ground, and can be represented by the final ground risk matrix R f indicating.

[0083] The final ground risk matrix R of the UAV f represents the ground risk impact on ground personnel when the UAV crashes on the ground, and it consists of two main parts, namely the initial ground risk matrix R of the UAV i and the effect matrix M of the ground risk mitigation measures of the UAV.

[0084] R f = R i ⊙M

[0085] Among them, the initial ground risk matrix R of the UAV i , that is, the inherent ground risk, is an inherent property of the UAV operation, and is only related to the characteristic parameters of the UAV (maximum characteristic size, maximum take-off mass, cruise speed) and the ground population density in the falling area.

[0086] R i (i, j, h) = D peo ·A cas

[0087] Wherein, D peo represents the ground personnel density in the UAV crash area, and A cas represents the casualty area area caused by the UAV crash.

[0088] M represents the effect of the ground risk mitigation measures, and is divided into two types, M1 and M2. The M1 type is to reduce the number of people directly exposed in the risk area, and can be represented by quantifying the protection effect of elements such as ground buildings on the personnel from being hit by the UAV.

[0089]

[0090] Wherein, P s is the ground shielding coefficient, and the parameter α is the impact energy value required for the fatality rate P s = 6, and the fatality rate P f = 50%. Generally, α = 34 kJ, and the parameter β is the energy threshold required for human fatality, generally taking β = 34 J. E crash is the kinetic energy at the time of UAV impact, and k is a correction factor, applicable to the case when the instantaneous kinetic energy of the UAV impact is relatively low, and can be calculated by the following formula:

[0091]

[0092] The M2 type is to reduce the impact caused by the UAV hitting the ground. Currently, a widely concerned method is to install a parachute on the UAV. By analyzing the kinematic process of the parachute system after the UAV fails in the air, the mitigation effect of the M2 measure is characterized.

[0093] Taking the multi-rotor UAV studied in the invention as an example, Figure 2 the falling process of the multi-rotor UAV equipped with a parachute after failure is analyzed. Under ideal conditions, affected by the gravity mg and the air resistance F airdrag it can be divided into two stages. In the figure, v represents the velocity direction, which is opposite to the direction of the air resistance, and t 1 is the opening time of the parachute system, and t 2 is the time when the UAV hits the ground. In the first stage, the UAV loses power and changes from the state of uniform horizontal flight to the ballistic descent state, and at this time the parachute system does not work; in the second stage, the parachute system opens and the UAV enters the parachute descent state.

[0094] During the falling process as Figure 2 shown, the three important variables affecting the risk to the ground are the horizontal velocity vertical velocity of the UAV when it hits the ground and the total falling time t 2 , and the calculation formulas are as follows:

[0095]

[0096] In the formula, m is the mass of the UAV, and respectively represent the instantaneous velocities of the UAV in the horizontal and vertical directions at time t 1 (the moment when the parachute system is fully deployed), ρ represents the air density, g is the acceleration due to gravity, and C d,p represents the air resistance coefficient of the whole parachute, and S p represents the effective windward area of the whole parachute.

[0097] The total falling time t 2 of the UAV is related to the height of the failure position and can be expressed as:

[0098]

[0099] In the formula, C d,u represents the air resistance coefficient of the UAV body, and S u represents the effective windward area of the UAV body.

[0100] In this embodiment, using the above calculation process, the three-dimensional ground risk matrix R of the UAV operating in the urban airspace can be calculatedn,m,l , and then calculate the equivalent safety level R(i, j, h).

[0101] Step S33: Update the longicorn beetle position according to the cost function value fed back by the antennae, select the direction with a lower cost for expansion, and update the longicorn beetle position to x t+1 , and generate an expansion node q bettle :

[0102] x t+1 = x t + δ t * d * sign[f(x r ) - f(x l )]

[0103] δ t+1 = δ t * η

[0104] In the formula, x t is the longicorn beetle position before update; δ t is the search step size of the longicorn beetle antenna algorithm at time t, and δ t+1 is the search step size of the longicorn beetle antenna algorithm at time t + 1; η is the reduction coefficient, d is the sensing distance of the longicorn beetle antenna angle; f(x r ) and f(x l ) are the cost functions corresponding to the search positions of the right antenna and the left antenna of the longicorn beetle respectively

[0105] Step S4: Use the collision detection function to detect obstacles between expansion nodes;

[0106] Specifically, connect the expansion node q bettle with the expansion node q new ; if there is no obstacle between the connected paths, add the expansion node q bettle to the search tree T, and connect it with the node q near , then execute S5; if there is an obstacle between the connected paths, return to execute S3;

[0107] Step S5: Build a tree, search for and output the optimal path;

[0108] Calculate the distance D between the expansion node q bettle and the end point q goal ; if the distance D is less than the threshold ε, output the optimal path; if the distance D is greater than the threshold ε, iteratively execute S2~S5 until the distance D is less than the threshold ε.

[0109] The RRT-BAS algorithm of this embodiment fully combines the random expansion ability of RRT and the direction selection ability of BAS. The addition of BAS makes the expansion process no longer completely random, but combines local optimization ability. Through path optimization and progressive optimization improvements, the quality of path planning is further improved. Through reconnection and neighborhood search, the RRT algorithm can continuously optimize the path and gradually converge to the global optimal solution; making the expansion path more in line with the trend of the optimal path and reducing the generation of redundant nodes.

[0110] To verify the effectiveness of the algorithm proposed in the present invention, RRT, BAS, RRT*, and RRT*-BAS are selected as baseline algorithms for comparative analysis, as Figure 4 , Figure 5 shown. The simulation results show that considering both path quality and convergence speed, the RRT-BAS algorithm of the present invention is significantly superior to the RRT, BAS, RRT*, and RRT*-BAS algorithms.

[0111] Although the specific implementation manners of the invention have been described in detail with reference to the accompanying drawings, it should not be construed as a limitation on the protection scope of this patent. Within the scope described in the claims, various modifications and deformations that can be made by those skilled in the art without creative efforts still fall within the protection scope of this patent.

Claims

1. A method for UAV path navigation planning based on urban low-altitude airspace risk constraints, characterized in that: The following steps are involved: S1. Set the parameters for the drone path navigation planning, including the search tree T and the starting point q of the drone path. start and the end point q goal ; S2, using the fast exploration random tree algorithm to randomly sample nodes q rand , and search for the node q in the search tree T rand The nearest node q near , and then expand the generated node q new ; S3, expand node q new As the beetle position, the beetle whisker algorithm is used to update the beetle position and generate the extended node q bettle ; S4. Connect to the extended node q bettle With the expansion node q new ; If there are no obstacles between the connected paths, the node q will be expanded bettle Add to the search tree T and compare with node q near Connect, then execute S5; if there is an obstacle between the connected paths, return to execute S3; S5. Calculate the extended node q bettle With the end point q goal The distance D between them; if the distance D is less than the threshold ε, the optimal path is output; if the distance D is greater than the threshold ε, S2 to S5 are iterated until the distance D is less than the threshold ε.

2. The method for unmanned aerial vehicle path navigation planning based on urban low-altitude airspace risk constraints according to claim 1 is characterized in that: The S2 specifically includes: Random sampling of nodes q in an urban airspace environment rand , search for the sampling node q in the search tree T rand The nearest node q near , if node q near With sampling node q rand If there is no obstacle on the line, the path search step Δq of the fast exploration random tree algorithm is used, starting from node q near To the sampling node q rand Expand in the direction to generate an extended node q new .

3. The method for unmanned aerial vehicle path navigation planning based on urban low-altitude airspace risk constraints according to claim 1 is characterized in that: The S3 comprises the following sub-steps: S31, expand node q new As the position of the beetle, the beetle whiskers are randomly generated and the random search unit vector is generated; S32, calculating the cost function of the search position of the two antennae of the longhorn beetle and the corresponding search position according to the longhorn beetle position and the longhorn beetle tentacles random search unit vector; S33, updating the position of the longicorn according to the cost function, thereby generating an extended node q bettle .

4. The method for unmanned aerial vehicle path navigation planning based on urban low-altitude airspace risk constraints according to claim 3 is characterized in that: In S31, the longicorn beetle randomly searches for a unit vector: Where b is the random search unit vector of the beetle whiskers; rnd(.) is a random vector, and k is the position dimension.

5. The method for UAV path navigation planning based on urban low-altitude airspace risk constraints according to claim 4 is characterized in that: The S32 specifically includes: Calculate the search position of the two antennae of the longicorn: In the formula, X r and X l The search positions of the right and left antennae of the longicorn respectively; d is the sensing distance of the longicorn whiskers; X is the extended node q new location; Calculate the cost function f(X) of the search position of the two antennae of the longicorn beetle: In the formula, α1 and α2 are the distance costs respectively. Risk Cost The weight coefficient of i , i=1,2,…,k, indicating that the drone is in RP i-1 RP i The comprehensive risk cost on the flight segment.

6. The method for unmanned aerial vehicle path navigation planning based on urban low-altitude airspace risk constraints according to claim 5 is characterized in that: Distance Cost and risk cost They are: in, In the formula, ||RP i-1 ,RP i ||For the UAV from the route node RP i-1 Fly to RP i The distance moved; v UAV is the cruising speed of the UAV; R(i,j,h) is the equivalent safety level of the UAV at the coordinate (i,j,h); R norm is the normalized risk level of the route in this area; Med(R(i,j,h)) is the median of all values ​​in the risk matrix of this segment.

7. According to the unmanned aerial vehicle path navigation planning method based on urban low-altitude airspace risk constraints according to claim 6, the equivalent safety level R (i, j, h) is specifically: R(i,j,h)=P(LOC)·R f (i,j,h) in, R f =R i ⊙M R i =D peo ·A cas Where P(LOC) is the probability that the drone fails and crashes while flying in the air, R f (i, j, h) is the final ground risk value of the drone crashing on the ground within the grid position (i, j, h); R f is the final ground risk value matrix; R i is the initial ground risk matrix; D peo is the ground personnel density in the area where the drone crashed, A cas is the area of ​​casualties caused by the crash of a drone; M is the effect matrix of ground risk mitigation measures, which includes two types: M1 and M2. Type M1 is to reduce the number of people directly exposed in the risk area, and type M2 is to reduce the impact caused by the drone hitting the ground.

8. According to the method for UAV path navigation planning based on urban low-altitude airspace risk constraints in claim 3, the position of the longhorn beetle is updated as x according to the cost function in S33 t+1 , and then generate the extended node q bettle , update the position of the longhorn beetle x t+1 for: x t+1 =x t +δ t *d*sign[f(x r )-f(x l )] d t+1 =d t *or In the formula, x t is the position of the longhorn beetle before updating; δ t is the search step length of the beetle beard algorithm at time t, δ t+1 is the search step length of the beetle whisker algorithm at time t+1; η is the reduction coefficient, and d is the sensing distance of the beetle whisker horn; f(x r ) and f(x l ) are the cost functions corresponding to the search position of the right antenna of the longicorn and the search position of the left antenna of the longicorn, respectively.

Images