A risk map-based unmanned aerial vehicle three-dimensional path planning method

Abstract

The application provides a kind of unmanned aerial vehicle three-dimensional path planning method based on risk map, comprising: discretizing low-altitude airspace based on rasterization technology;Acquire urban airspace building data and set unmanned aerial vehicle geographic fence;According to the risk assessment model of unmanned aerial vehicle to the ground risk;Generate a risk map based on probability;Establish a path planning model based on risk map;Through the jump point search algorithm, the three-dimensional path planning of unmanned aerial vehicle in complex environment is carried out.The city low-altitude airspace is taken as the research object, by combining the risk map and the path planning algorithm, the strategic stage of unmanned aerial vehicle autonomous path planning can be carried out in the complex urban environment with multiple factors coexisting, the reasonable static collision avoidance is realized while considering the global optimization, the potential risk of casualties caused by unmanned aerial vehicle operation to ground pedestrians is effectively reduced, to realize the safe operation of unmanned aerial vehicle.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicles (UAVs), and more specifically to a UAV three-dimensional path planning method based on a risk map. Background Technology

[0002] With the continuous development of drone technology and the rapid advancement of its application prospects, the flexibility and high maneuverability of drones have been fully demonstrated in urban airspace, making urban air mobility (UAM) possible. However, in the complex environment of urban low-altitude airspace, such as CBD (Central Business District) areas, how to strategically plan a safe, collision-free, and as short as possible flight path to ensure the safe operation of drones is one of the major problems that urgently needs to be solved.

[0003] Currently, research on UAVs for urban airspace management focuses on conflict resolution and scenario applications, aiming to achieve safe operation under idealized scenarios. UAV path planning is one of the key research areas, with algorithms mainly falling into two categories: graph-based search methods and optimal control-based methods. Graph-based search methods often face the curse of dimensionality, while optimal control-based methods typically rely on numerical solutions and face the minima problem. In 2011, Adolf FM and Andert.F. published "Rapid multi-query path planning for a vertical take-off and landing unmanned aerial vehicle," proposing an online multi-query path planning method that combines sampling-based motion planning with real-time path search to achieve overall maneuverability in scenarios with unpredictable obstacle changes, thereby improving the autonomy of UAV operation. In 2013, Wang Yi published "UAV Path Planning Algorithm Based on PH Curve," proposing a method to directly use the curvature-continuous PH curve for UAV path planning. Utilizing the continuous curvature, smoothness, and rationality properties of the PH curve, and combining the advantages of genetic algorithms and simulated annealing algorithms, the method enhances global search capabilities, providing a basis for UAV flight control. In 2020, Li Xianqiang published "Improved Design of Ant Colony Algorithm and Its Application in Path Planning". The paper designed a new ant colony algorithm that can avoid the problem of premature convergence and getting trapped in local optima in the traditional ant colony algorithm. Compared with the traditional ant colony algorithm, it improves the convergence speed. By dividing the three-dimensional space into grids, the optimization method was successfully applied to the three-dimensional path planning of UAVs.

[0004] It is clear that the above research on UAV path planning focuses primarily on ideal environments, failing to consider the significant risks posed by the complexity of obstacle shapes and distribution. Deviations between the UAV path and the actual operating scenario can lead to conflicts. Furthermore, traditional 3D path planning algorithms do not account for the risk of injury or death to pedestrians on the ground should a UAV malfunction, thus failing to adequately meet the requirements for safe operation. Summary of the Invention

[0005] To address the aforementioned technical shortcomings, the purpose of this invention is to provide a three-dimensional path planning method for unmanned aerial vehicles (UAVs) based on risk maps.

[0006] The present invention adopts the following technical solution:

[0007] A method for UAV 3D path planning based on a risk map includes the following steps:

[0008] Step 1: Discretize the low-altitude airspace;

[0009] Step 2: Obtain urban airspace building data and delineate no-entry geofences for drones;

[0010] Step 3: Establish a risk assessment model based on the ground risks posed by drones;

[0011] Step 4: Generate a risk map based on the probability of injury or death;

[0012] Step 5: Establish a path planning model based on the risk map;

[0013] Step 6: Perform 3D path planning for the UAV.

[0014] In step one, the low-altitude airspace is discretized using a rasterization technique.

[0015] The urban airspace building data in step two includes at least one of the following: static buildings, UAV flight performance, airspace information, and population density information.

[0016] Based on the acquired urban airspace building data, geofences are drawn to delineate no-fly zones, temporary take-off and landing points, and the locations of obstacles to prevent drones from entering.

[0017] The risk assessment model in step three is as follows:

[0018] Determine the area of ​​the ground affected by a collision based on a model of the drone's descent type;

[0019] Drone accidents causing injuries or fatalities can be categorized into three scenarios: a drone descends out of control and crashes into a person on the ground, resulting in death; the following formula can be used:

[0020]

[0021] Among them, ρ(x,y) refers to the population density at the geographical location (x,y), A exp P refers to the area of ​​the ground affected by the collision, S(x,y) represents the shading coefficient at the geographical location (x,y), S(x,y)∈[0,10], E[·] refers to the expected value, and E(x,y) represents the kinetic energy of the UAV impacting the geographical location (x,y); casualty (x,y) represents the probability of injury or death; P decline P represents the probability of an uncontrolled descent of a drone. impact (x,y) represents the probability of a drone colliding with a person on the ground; P fatality (x,y) represents the probability of death caused by a drone crash; β is the impact kinetic energy threshold for death caused by a drone crash when S(x,y)=0, which is taken as 34J according to the definition of the death limit; α is the impact kinetic energy threshold for a 50% probability of death caused by a drone crash when S(x,y)=6, which is taken as 10J. 6 J;

[0022] Casualties are defined using the following formula:

[0023]

[0024] Among them, P casualty (x,y) represents the probability of injury or death; The probability of casualties corresponding to a ballistic descent; The probability of injury or death corresponding to the occurrence of loss of control and skidding.

[0025] The model for the drone's descent type includes a ballistic descent mode and a runaway glide descent mode, and the impact area for the two descent modes are as follows:

[0026] A exp1 =1.15·π(r) UAV +r p ) 2 ;

[0027] A exp2 =2·(r UAV +2r p )·d+π(r UAV +2r p ) 2 ;

[0028] in, r UAV Let r be the radius of the circumscribed sphere of the drone. p h is the average width of a human body pThe average human height is taken as 1.7m; d is the horizontal distance the drone travels after landing at pedestrian height; θ is the gliding angle, representing the angle between the velocity vector and the ground.

[0029] The risk map generated in step four uses the following method:

[0030] The probability values ​​of the risk assessment model in step three are statistically analyzed to generate a casualty probability map;

[0031] Based on obstacle data, population density data, and airspace information, a corresponding airspace population density layer, obstacle layer, and occlusion factor layer are generated.

[0032] The risk map is obtained by merging the casualty probability map with the population density layer, barrier layer, and shading factor layer.

[0033] The path planning model in step five includes:

[0034] Design heuristic functions and constraints based on grid technology;

[0035] The element values ​​of the risk map are incorporated into the heuristic function to update the constraints.

[0036] The path planning model is as follows:

[0037]

[0038] Where α and β are weighting coefficients; Average U P represents the average probability of injury or death per flight hour for the unmanned aerial vehicle (U) under the current environment. casualty (x C ,y C p represents the probability of injury or death per flight hour for pedestrians at the current location; reward f(r) represents the reward function value corresponding to the probability of injury or death. C ) is a heuristic function; g(r) C (r) represents the actual cost from the initial position to the current position; C x is the estimated cost of the remaining path from the current location to the target location; i ,y i ,z i x represents the coordinates of the previous position. i+1 ,y i+1 ,z i+1 x represents the coordinates of the next position. C ,y C ,z C x is the coordinate of the current jump point; G ,y G ,z G ω represents the target location coordinates; υ represents the reward parameters.

[0039] The UAV 3D path planning in step six adopts the following method:

[0040] A jump point search algorithm is used for 3D path planning to obtain an initial reference path;

[0041] Instead of path dediagonalization, a method of delineating UAV geofences for various static obstacles and restricted areas is used to optimize the reference path and obtain the final UAV 3D path.

[0042] The beneficial effects of this invention are as follows: This invention takes urban low-altitude airspace as the research object. By combining risk maps and path planning algorithms, it can perform autonomous path planning for UAVs in the strategic stage in complex urban environments where multiple factors coexist. While considering global optimization, it can achieve reasonable static collision avoidance, effectively reducing the potential injury and death risks to pedestrians on the ground caused by UAV operation, so as to achieve safe operation of UAVs. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a flowchart of the UAV 3D path planning method based on risk map provided by the present invention.

[0045] Figure 2 This is a schematic diagram of the impact zone in the ballistic descent mode.

[0046] Figure 3 This is a schematic diagram of the impact area during an uncontrolled gliding descent.

[0047] Figure 4 This is a schematic diagram of a risk map.

[0048] Figure 5 A schematic diagram of "forced neighbor" and "jump point" defined for the jump point search algorithm.

[0049] Figure 6 This represents a theoretical extension of the jump point search algorithm.

[0050] Figure 7 This is an extension method used in the actual execution of the jump point search algorithm. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] like Figures 1 to 7 As shown, this invention provides a three-dimensional path planning method for unmanned aerial vehicles (UAVs) based on a risk map, comprising the following steps:

[0053] Step 1: Discretize the low-altitude airspace;

[0054] Step 2: Obtain urban airspace building data and delineate no-entry geofences for drones;

[0055] Step 3: Establish a risk assessment model based on the ground risks posed by drones;

[0056] Step 4: Generate a risk map based on the probability of injury or death;

[0057] Step 5: Establish a path planning model based on the risk map;

[0058] Step Six: Perform 3D Path Planning for the UAV

[0059] In step one above, the low-altitude airspace is discretized using a rasterization technique;

[0060] By selecting an appropriate raster granularity to discretize the spatial domain, in the three-dimensional world R(X×Y×Z×l), each raster can be represented as r(x,y,z), where l is the raster granularity, x∈X, y∈Y, z∈Z, and

[0061] The urban airspace building data in step two should include at least one of the following: static buildings, UAV flight performance, airspace information, and population density information.

[0062] Based on the acquired urban airspace building data, appropriate geofencing buffer zones are selected, and no-fly zones, temporary take-off and landing points, and locations of obstacles are designated as drone-restricted geofencing areas.

[0063] The risk assessment model in step three is as follows:

[0064] Determine the area of ​​the ground affected by a collision based on a model of the drone's descent type;

[0065] Drone descent types include ballistic descent and runaway glide; the corresponding drone impact area can be calculated based on the different types (e.g., Figure 2 and Figure 3 (as shown);

[0066] The impact area areas are: A exp1 =1.15·π(r) UAV +r p ) 2 A exp2 =2·(r UAV +2r p )·d+π(r UAV +2r p ) 2 ;

[0067] in, r UAV Let r be the radius of the circumscribed sphere of the drone. p h is the average width of a human body p The average human height is taken as 1.7m (this value can be set according to the actual situation), d is the horizontal distance the drone moves after landing at the height of a pedestrian; θ is the gliding angle, which represents the angle formed by the velocity vector and the ground;

[0068] Drone accidents causing injuries or fatalities can be categorized into three scenarios: a drone descends out of control and crashes into a person on the ground, resulting in death; the following formula can be used:

[0069]

[0070] Among them, j is a general parameter; ρ(x,y) refers to the population density at the geographical location (x,y), A exp P refers to the area of ​​the ground affected by the collision, S(x,y) represents the shading coefficient at the geographical location (x,y), S(x,y)∈[0,10], E[·] refers to the expected value, and E(x,y) represents the kinetic energy of the UAV impacting the geographical location (x,y); casualty (x,y) represents the probability of injury or death; P decline P represents the probability of an uncontrolled descent of a drone. impact (x,y) represents the probability of a drone colliding with a person on the ground; P fatality (x,y) represents the probability of death caused by a drone crash; β is the impact kinetic energy threshold for death caused by a drone crash when S(x,y)=0, which is taken as 34J according to the definition of the death limit; α is the impact kinetic energy threshold for a 50% probability of death caused by a drone crash when S(x,y)=6, which is taken as 10J. 6 J;

[0071] Casualties are defined using the following formula:

[0072]

[0073] Among them, Pcasualty (x,y) represents the probability of injury or death; The probability of casualties corresponding to a ballistic descent; The probability of injury or death corresponding to the occurrence of loss of control and skidding.

[0074] Step four uses the following method to generate the risk map:

[0075] The probability values ​​of the risk assessment model in step three are statistically analyzed to generate a casualty probability map;

[0076] Based on obstacle data, population density data, and airspace information, a corresponding airspace population density layer, obstacle layer, and occlusion factor layer are generated.

[0077] By merging the casualty probability map with the population density layer, barrier layer, and shading factor layer, a risk map is obtained; for example... Figure 4 As shown, the element values ​​in the risk map are defined to quantify the risks that drones pose to the ground environment (mainly people).

[0078] The path planning model in step five is designed as follows:

[0079] Design a grid-based heuristic function and constraints, incorporate risk map element values ​​into the heuristic function, and update the constraints. The path planning model based on risk paths is as follows:

[0080] min f(r C )=α·(g(r C )+(r C ))+β·p reward

[0081]

[0082] Where α and β are weighting coefficients; Average U P represents the average probability of injury or death per flight hour for the unmanned aerial vehicle (U) under the current environment. casualty (x C ,y C p represents the probability of injury or death per flight hour for pedestrians at the current location; feward f(r) represents the reward function value corresponding to the probability of injury or death. C ) is a heuristic function; g(r) C (r) represents the actual cost from the initial position to the current position; C x is the estimated cost of the remaining path from the current location to the target location; i ,y i ,z i x represents the coordinates of the previous position. i+1 ,y i+1 ,z i+1x represents the coordinates of the next position. C ,y C ,z C x is the coordinate of the current jump point; G ,y G ,z G ω and υ are the target location coordinates; ω and υ are the reward parameters.

[0083] The UAV 3D path planning in step six adopts the following method:

[0084] A jump point search algorithm is used for 3D path planning of the UAV to obtain an initial reference path, such as... Figure 5 As shown, the jump point search algorithm calculates as few heuristic function points as possible by defining "forced neighbors" and "jump points" to speed up pathfinding time in a given map. The algorithm's expansion direction in the rasterized spatial domain and the expansion methods during actual algorithm execution are as follows: Figure 6 and Figure 7 As shown, the method of using UAV geofences to delineate various static obstacles and restricted areas replaces the path dediagonalization method to optimize the reference path and improve pathfinding efficiency.

[0085] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for UAV 3D path planning based on a risk map, characterized in that, Includes the following steps: Step 1: Discretize the low-altitude airspace; Step 2: Obtain urban airspace building data and delineate no-entry geofences for drones; Step 3: Establish a risk assessment model based on the ground risks posed by drones; Step 4: Generate a risk map based on the probability of injury or death; Step 5: Establish a path planning model based on the risk map; Step Six: Perform 3D path planning for the UAV; The risk assessment model in step three is as follows: Determine the area of ​​the ground affected by a collision based on a model of the drone's descent type; Drone accidents causing injuries or fatalities can be categorized into three scenarios: a drone descends out of control and crashes into a person on the ground, resulting in death; the following formula can be used: Among them, ; Geographical location Population density at the location It refers to the area of ​​the ground affected by a collision. Indicates geographical location The shading coefficient at that location, , The expected value. Indicates the geographical location of the drone impact. Kinetic energy at the point; Probability of injury or death; The probability of an uncontrolled descent of a drone; The probability of a drone colliding with a person on the ground; The probability of a person dying from a drone crash; for At that time, the impact kinetic energy threshold for death caused by a drone crash is set at 34J according to the definition of the death limit; for At that time, the impact kinetic energy threshold at which a drone crash results in a 50% probability of death is set to 10. 6 J; Casualties are defined using the following formula: in, Probability of injury or death; The probability of casualties corresponding to a ballistic descent; The probability of injury or death corresponding to the occurrence of loss of control and skidding.

2. The UAV 3D path planning method based on a risk map as described in claim 1, characterized in that, In step one, the low-altitude airspace is discretized using a rasterization technique.

3. The UAV 3D path planning method based on a risk map as described in claim 1, characterized in that, The urban airspace building data in step two includes at least one of the following: static buildings, UAV flight performance, airspace information, and population density information. Based on the acquired urban airspace building data, geofences are drawn to delineate no-fly zones, temporary take-off and landing points, and the locations of obstacles to prevent drones from entering.

4. The UAV 3D path planning method based on a risk map as described in claim 1, characterized in that, The model for the drone's descent type includes a ballistic descent mode and a runaway glide descent mode, and the impact area for the two descent modes are as follows: ； ； in, , , The radius of the circumscribed sphere for the drone size. The average width of a human body. The average human height is taken as 1.7m. This refers to the horizontal distance the drone travels after landing at pedestrian height. The gliding angle represents the angle between the velocity vector and the ground.

5. The UAV 3D path planning method based on a risk map as described in claim 1, characterized in that, The risk map generated in step four uses the following method: The probability values ​​of the risk assessment model in step three are statistically analyzed to generate a casualty probability map; Based on obstacle data, population density data, and airspace information, a corresponding airspace population density layer, obstacle layer, and occlusion factor layer are generated. The risk map is obtained by merging the casualty probability map with the population density layer, barrier layer, and shading factor layer.

6. The UAV 3D path planning method based on a risk map as described in claim 1, characterized in that, The path planning model in step five includes: Design heuristic functions and constraints based on grid technology; The element values ​​of the risk map are incorporated into the heuristic function to update the constraints. The path planning model is as follows: in, , These are the weighting coefficients; For drones The average probability of injury or death per flight hour under the current conditions; This represents the probability of injury or death per flight hour for pedestrians at the current location. The reward function value corresponding to the probability of injury or death; It is a heuristic function; The actual cost from the initial position to the current position; Estimate the cost of the remaining path from the current location to the target location; The coordinates of the previous position; These are the coordinates of the next position; The coordinates of the current jump point; The target location coordinates; For reward parameters.

7. The UAV 3D path planning method based on a risk map as described in claim 1, characterized in that, The UAV 3D path planning in step six adopts the following method: A jump point search algorithm is used for 3D path planning to obtain an initial reference path; Instead of path dediagonalization, a method of delineating UAV geofences for various static obstacles and restricted areas is used to optimize the reference path and obtain the final UAV 3D path.

Images