Unmanned aerial vehicle path planning method based on ellipse tangent graph method

Abstract

The invention relates to the technical field of unmanned aerial vehicle control, in particular to an unmanned aerial vehicle path planning method based on an elliptical tangent graph method, and the method comprises the steps: a flight environment modeling step, which is used for building an elliptical obstacle model with a building as an obstacle, the method comprises a three-dimensional flight environment model construction step for constructing a three-dimensional flight environment model based on elevation information of a flight area and distribution of an obstacle model, a three-dimensional path acquisition step for performing path search to output a flight path, and a three-dimensional path smoothing step for smoothing path points of the flight path output by the three-dimensional path acquisition step, and obtaining a flight path for unmanned aerial vehicle control. According to the method, energy consumption is taken as a core optimization target of path planning, so that the problem of comprehensive decision-making among path length, motion smoothness and energy consumption in a traditional method is effectively solved, and the task range of the unmanned aerial vehicle is remarkably extended.

Description

Technical Field

[0001] The present application relates to the field of drone control technology, and in particular to a drone path planning method based on an ellipse tangent graph method. Background Art

[0002] Unmanned aerial vehicles (UAVs) are widely used in various fields, including traffic inspection, disaster relief, cargo transportation, and target reconnaissance. Path planning, as the core component of autonomous control for UAV systems, is essential for ensuring efficient mission execution and plays a key role in advancing UAV applications.

[0003] Among existing graph-based path planning methods, the Voronoi diagram method improves planning efficiency by dividing free space into multiple convex polygonal regions. Its advantages lie in its relatively low computational complexity and ability to ensure that paths maintain the maximum safe distance from obstacles. However, this axis-based planning approach often results in generated path lengths significantly deviating from the theoretical optimal value, making it poorly suited for applications requiring precise paths. In contrast, the A* algorithm, with its heuristic search strategy, demonstrates superior performance in complex environments. By comprehensively evaluating path costs and heuristic functions, it can effectively find optimal or near-optimal paths. However, this algorithm requires maintaining open and closed lists when expanding nodes. As the environment becomes more complex, this algorithm faces explosive node growth and excessive memory usage. Strategies such as hierarchical search and skip-point optimization are often employed to improve performance. Another classic approach, the tangent graph method, employs a more intuitive planning approach, constructing a path network by calculating common tangents between obstacle polygons. This method is simple to implement and ensures path connectivity, but the generated paths often suffer from excessive tortuosity and proximity to obstacles. Although researchers have proposed using circular or elliptical expansion models to smooth the path, these improvements are currently only applicable to two-dimensional space and cannot meet the planning requirements of three-dimensional moving objects such as drones.

[0004] In addition, existing intelligent optimization algorithms include GA, ACO and PSO. These methods have global optimization capabilities, but they take a long time in complex environments and are prone to falling into local optimality.

[0005] The flight endurance of drones is limited by battery capacity and power efficiency. Improper energy management can lead to mission interruptions, equipment crashes, or data loss. However, in practical applications, existing research on drone path planning primarily focuses on obstacle avoidance, path length, and planning time. Power consumption is often overlooked, making it difficult to ensure drone flight endurance during mission execution. Summary of the Invention

[0006] An embodiment of the present application provides a UAV path planning method based on the elliptical tangent graph method. By taking energy consumption as the core optimization target of path planning, it effectively solves the comprehensive decision-making problem of traditional methods between path length, motion smoothness and energy consumption, and significantly extends the mission range of the UAV.

[0007] To achieve the above objectives, the present invention provides a method for UAV path planning based on the ellipse tangent graph method, comprising: The flight environment modeling step involves determining the flight area, obtaining the geographic location information, size and structure information of buildings within the flight area, constructing an elliptical obstacle model using the buildings as obstacles, and constructing a three-dimensional flight environment model based on the elevation information of the flight area and the distribution of the obstacle model. A three-dimensional path acquisition step is to connect the path ST from the starting point S to the end point T, and perform collision detection on the path ST based on the three-dimensional flight environment model. If no collision occurs, the drone is controlled to fly along the path ST; If a collision occurs, the path point acquisition step is executed to generate a two-dimensional reference path point and determine whether the height of the starting point S and the end point T are the same; If they are the same, a flight path to bypass the obstacle is generated based on the bypass strategy or the climb strategy; If they are different, an alternative path is generated and the flight path with the lowest consumption among the alternative paths is output based on a flight energy consumption model, wherein n path points between the starting point S and the end point T are obtained based on the two-dimensional reference path point, and each path point is configured with a detour strategy and a climb strategy to generate Alternative paths are generated, and the UAV flight action is decomposed into take-off phase, landing phase, and horizontal flight phase according to the alternative paths. Based on the energy consumption of the take-off phase, Energy consumption during landing phase , energy consumption during horizontal flight A flight energy consumption model is constructed, and the flight energy consumption model is used to output the flight path with the lowest energy consumption among the alternative paths. The flight energy consumption model is represented by the following calculation model: ; The three-dimensional path smoothing step smoothes the path points of the flight path output by the three-dimensional path acquisition step to obtain a flight trajectory for controlling the UAV.

[0008] In some embodiments, the energy consumption during takeoff is Calculated based on the following calculation model: , in, is the ground height, is the flight altitude of the UAV, represents the vertical upward velocity component, , , Respectively represent the propeller working efficiency, motor working efficiency and electronic speed regulator working efficiency, is the total mass of the UAV, is the acceleration due to gravity, , Indicates the air density, the value is 1.225 ; is the flight time, is the temperature gradient below the top of the troposphere, with a value of -0.0065 , R is the gas constant in air, which is 287.05287 , is the propeller disc area, is the correction factor, is the standard temperature, which is 288.12K, or 15°C.

[0009] The above calculation model is based on the energy consumption modeling of the UAV during the process of taking off from the ground to the flight altitude. The energy required for the UAV to climb from the ground to the target altitude is converted into an integral expression. The various actual influencing factors such as the change of air density with altitude, propulsion system efficiency, flight speed, etc. are considered. The temperature change term is Used to simulate the standard atmospheric temperature gradient, indicating the impact of the environment on air density and lift, efficiency parameter The loss of the UAV propulsion system is corrected, and the instantaneous power consumption of the entire process is calculated by integration to obtain the total energy consumption estimation result of the take-off phase, so as to achieve a more accurate estimation of the energy demand of the UAV during the vertical take-off phase and improve the authenticity and adaptability of the energy consumption prediction.

[0010] In the above embodiment, if the ground temperature distribution and wind speed model are known to be relatively complex, a piecewise temperature function may be introduced or fitted based on measured data to replace the standard temperature gradient term.

[0011] In the above embodiment, the propeller operating efficiency, the motor operating efficiency, and the electronic speed regulator operating efficiency can be obtained through actual flight tests or characteristic curves provided by the manufacturer, or can be set as typical constants for approximate estimation when accurate values are not available.

[0012] In some embodiments, the energy consumption during the landing phase Calculated based on the following calculation model: , in, represent the vertical downward velocity components respectively.

[0013] The above calculation model uses integration to represent the instantaneous power accumulation of the UAV during the process of descending from the target flight altitude to the ground. The temperature change term is the same as the take-off phase, which is used to represent the temperature change as the altitude decreases over time. Compared with the traditional static estimation method, this model is more in line with the actual aerodynamic characteristics and can be used to optimize landing posture and path selection, thereby reducing unnecessary energy consumption and system load, and improving system energy-saving efficiency and flight safety.

[0014] In some embodiments, the energy consumption during the horizontal flight phase is Calculated based on the following calculation model: , in, is the horizontal flight distance, is the horizontal velocity component, is the windward area; is the zero drag lift coefficient; is the induced drag factor, , Indicates the air density at the current path point height The air density at the next path point The average value of the air density at the current path point height The air density at the next path point It can be calculated based on the following calculation model: .

[0015] The above calculation model models horizontal flight energy consumption into two components: air resistance power consumption, which is the energy consumed per unit time by the drone to counteract air resistance in the forward direction; and lift power consumption, which is the power consumed to maintain lift against gravity. These two components are used to model the energy consumption during horizontal flight, and then corrected using the propulsion system efficiency as the denominator. This fully simulates the actual energy requirements of the drone during complex flight, helping to prioritize and optimize the energy efficiency of different horizontal segments. Compared to traditional simplified models that only estimate drag energy consumption, this implementation is more practical in engineering and is particularly suitable for path planning for long-distance, variable-altitude missions, enhancing the system's ability to refine energy budgets.

[0016] In the above embodiment, the windward area , zero drag lift coefficient , Induced drag factor When actual measurements are not possible, the parameters can be estimated based on the UAV platform combined with the model and Fengdong measurement standard parameters, or approximate constants can be used in combination with the flight scenario to fine-tune the configuration.

[0017] In some embodiments, the path point acquisition step includes: The three-dimensional flight environment model is projected into a two-dimensional plane obstacle map, and the two-dimensional reference path points from the starting point S to the end point T are generated using the improved ellipse tangent map method.

[0018] In some embodiments, the path point acquisition step further comprises: In the path initialization step, a path node set, a candidate node queue, and a collision obstacle set are initialized to empty sets. The path node set is represented as Path and is used to record valid nodes in the current path. The candidate node queue is represented as C and is used to temporarily store path nodes to be expanded. The collision obstacle set is represented as B and is used to record obstacles that collide with obstacles during the path search. In a bidirectional path search step, the starting point S is added to the path node set Path, and the end point T is added to the candidate node queue C as the preset end point. A path collision detection step is performed to generate a first candidate path set from the starting point S to the end point T. The end point T is then added to the path node set Path, and the starting point S is added to the candidate node queue C as the preset end point. A path collision detection step is performed to iteratively generate a second candidate path set from the end point T to the starting point S. A bidirectional path topology graph is established based on the first and second candidate path sets, and collision-free path nodes and their connection relationships are recorded. an optimal path screening step, combining the first candidate path set and the second candidate path set, calculating the total length of each path using a path cost matrix, selecting a collision-free path with the minimum total length in the bidirectional path topology graph using the Dijkstra algorithm, and outputting a two-dimensional reference path point of the global shortest path; Among them, during the collision detection process, the tangent graph method is used to generate candidate temporary paths, and a temporary path from the candidate temporary paths is selected for path generation based on preset heuristic rules to ensure that the path detours between obstacles reasonably and has high planning efficiency, avoiding falling into local optimality.

[0019] In some embodiments, the path collision detection step includes: The end node O in the path node set Path is obtained, and the end node in the candidate node queue C is obtained as the target node D. The end node O and the target node D are connected to obtain the path segment OD for collision detection. If a collision is detected, the path is replanned until the path reaches the preset end point without collision with obstacles, thus generating a candidate path. When performing reverse iteration, the preset end point represents the starting point S.

[0020] When the path collision detection step is used to perform forward and reverse iterations, the path finding quality in a complex obstacle-intensive environment can be improved.

[0021] In some embodiments, the path collision detection step further comprises: Obtain the end node O in the path node set Path; Perform collision detection on the path segment OD. If it does not intersect with the obstacle, move the target node D from the candidate node queue C to the path node set Path. Otherwise, Determine the first collision obstacle Obs, obtain the first and second tangents between the end node O and the collision obstacle Obs, and obtain the third and fourth tangents between the end point T and the collision obstacle Obs. Generate two candidate temporary paths based on the intersection point N1 of the first and third tangents and the intersection point N1′ of the second and fourth tangents. Based on a preset heuristic rule, an optimal intersection point N* is determined between the intersection point N1 and the intersection point N1′ as a relay point of the current candidate path; Connect the optimal intersection point N* and the end node O for collision detection. If there is no intersection with the obstacle, add the optimal intersection point N* to the path node set Path and use the optimal intersection point N* as the new end node O for loop iteration. Otherwise, add the optimal intersection point N* to the candidate node queue C and add the collision obstacle Obs to the collision obstacle set B. If the candidate node queue C is not empty, obtain the target node D′ in the candidate node queue C, repeat the above steps, and connect the end node O and the target node D′ for collision detection.

[0022] In some embodiments, in the flight environment modeling step, the top view outer contour of the building is modeled as an ellipse to obtain an obstacle model to represent the restricted area for drone flight. The obstacle model is represented as follows: , In the above formula, and Respectively represent the semi-major axis and semi-minor axis of the ellipse; is the inclination angle between the semi-major axis and the positive direction of the horizontal axis in the two-dimensional obstacle map; Indicates the safe distance between the drone and obstacles; Indicates the center coordinates of the obstacle; Indicates the height of the obstacle.

[0023] Based on the obstacle model and the actual geometric boundary characteristics of the obstacle in the flight area, the modeling efficiency and path algorithm compatibility are improved by approximating it to an ellipse. In three-dimensional space, the obstacle area is a closed elliptical cylinder, which restricts the drone's flight path, simplifies the obstacle representation in the flight environment, and improves modeling speed and algorithm parsing efficiency. This modeling method is highly parameterized and can achieve approximate representation of the diverse shapes of buildings. By configuring safety distances, the no-fly boundary can be dynamically adjusted to implement flight safety strategies that adapt to different mission risk levels. Compared to polygonal or voxel modeling methods, this solution has low computational overhead, continuous representation, and facilitates geometric operations and shortest path searches in path planning algorithms.

[0024] In some embodiments, the three-dimensional path smoothing step uses a cubic B-spline curve for smoothing.

[0025] Compared with related technologies, the UAV path planning method based on the elliptic tangent graph method provided in the embodiment of the present application extends the elliptic tangent graph method to three-dimensional scenes, converts building obstacles into elliptical obstacle models, enhances adaptability to irregular building structures, and realizes effective acquisition of two-dimensional reference paths. By combining obstacle heights and two-dimensional reference paths, modeling and path search of complex three-dimensional flight environments are realized, and the energy consumption is calculated by the energy consumption model to select the solution with the lowest energy consumption, thereby enhancing the flexibility and adaptability of path planning. It is especially suitable for typical urban air traffic scenarios such as urban canyons and buildings with height differences or multi-level obstacles to avoid path collision problems.

[0026] The details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 is a flow chart of a method for planning a path for a drone according to an embodiment of the present application; Figure 2 is a flowchart of a three-dimensional path acquisition step according to an embodiment of the present application; Figure 3 is a flowchart of the path point acquisition steps according to an embodiment of the present application; Figure 4 A two-dimensional obstacle map showing the distribution of low-density obstacles; Figure 5 It is another two-dimensional obstacle map with low-density obstacle distribution; Figure 6A two-dimensional obstacle map with high-density obstacle distribution; Figure 7 This is a schematic diagram of the path planning comparison effect under C1 distribution; Figure 8 This is a schematic diagram of the path planning comparison effect under C2 distribution; Figure 9 This is a schematic diagram of the path planning comparison effect under C2 distribution; Figure 10 for Figure 7 Schematic diagram of the overhead effect; Figure 11 for Figure 8 Schematic diagram of the overhead effect; Figure 12 for Figure 9 Schematic diagram of the overhead effect. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solutions and advantages of this application more clearly understood, the present application is described and illustrated below in conjunction with the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely used to explain this application and are not intended to limit this application. Based on the embodiments provided in this application, all other embodiments obtained by those of ordinary skill in the art without making any creative efforts are within the scope of protection of this application.

[0029] Obviously, the drawings described below are merely examples or embodiments of the present application. Those skilled in the art can, without inventive effort, apply the present application to other similar scenarios based on these drawings. Furthermore, it is also understood that, although the effort involved in such a development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, changes in design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as an insufficiency of the content disclosed in this application.

[0030] The present invention provides a method for UAV path planning based on the ellipse tangent graph method. Figures 1 to 3 is a flow chart of a method for planning a drone path according to an embodiment of the present application. Figures 1 to 3 As shown, the process includes the following steps: The flight environment modeling step S1 determines the flight area, obtains the geographical location information, size and structure information of the buildings in the flight area, constructs an elliptical obstacle model using the buildings as obstacles, and constructs a three-dimensional flight environment model based on the elevation information of the flight area and the distribution of the obstacle model. In the three-dimensional path acquisition step S2, a path ST is formed connecting the starting point S to the end point T, and collision detection is performed on the path ST based on the three-dimensional flight environment model. If no collision occurs, the drone is controlled to fly along the path ST; If a collision occurs, the path point acquisition step S21 is executed to generate a two-dimensional reference path point and determine whether the heights of the starting point S and the end point T are the same; If they are the same, a flight path to bypass the obstacle is generated based on the bypass strategy or the climb strategy; If they are different, an alternative path is generated and the flight path with the lowest consumption among the alternative paths is output based on a flight energy consumption model, wherein n path points between the starting point S and the end point T are obtained based on the two-dimensional reference path point, and each path point is configured with a detour strategy and a climb strategy to generate Alternative paths are generated, and the UAV flight action is decomposed into take-off phase, landing phase, and horizontal flight phase according to the alternative paths. Based on the energy consumption of the take-off phase, Energy consumption during landing phase , energy consumption during horizontal flight constructing a flight energy consumption model, and using the flight energy consumption model to output a flight path with the lowest energy consumption among the alternative paths; The three-dimensional path smoothing step S3 smoothes the path points of the flight path output by the three-dimensional path acquisition step to obtain a flight trajectory for drone control, improves the smoothness of the drone flight trajectory, eliminates sharp turns in the path, ensures flight stability, and reduces sharp turns and jitters during flight.

[0031] In the above embodiment, the detour strategy is: based on the two-dimensional reference path, detour from the current path point to the next path point, mark the next path point as the new starting point and update it to the starting point S, and then execute the path point acquisition step S21; the obstacle avoidance strategy is: climb from the current path point to above the obstacle, mark the top of the obstacle as the new starting point and update it to the starting point S, and then execute the path point acquisition step.

[0032] Based on the above steps, this application extends the ellipse tangent graph method to three-dimensional scenes, transforms building obstacles into ellipse obstacle models, enhances adaptability to irregular building structures, and realizes effective acquisition of two-dimensional reference paths. By combining obstacle height and two-dimensional reference paths, the modeling and path search of complex three-dimensional flight environments are realized. In the path search process, detour strategies and climb strategies are introduced to determine whether to fly directly or perform path adjustment. The path point combination logic is used to combine n path points into a The system selects the solution with the lowest energy consumption by calculating the energy consumption through an energy consumption model, enhancing the flexibility and adaptability of path planning. It is particularly suitable for typical urban air traffic scenarios such as urban canyons and buildings with height differences or multi-level obstacles, avoiding path collision problems.

[0033] In the above embodiment, the flight energy consumption model is represented by the following calculation model: , which improves the autonomous decision-making and path planning capabilities of drones in complex environments; , , , in, is the ground height, is the flight altitude of the UAV, Represents the vertical upward velocity component represent the vertical downward velocity components, , , Respectively represent the propeller working efficiency, motor working efficiency and electronic speed regulator working efficiency, is the total mass of the UAV, is the acceleration due to gravity, , Indicates the air density, the value is 1.225 ; is the flight time, is the temperature gradient below the top of the troposphere, with a value of -0.0065 , R is the gas constant in air, which is 287.05287 , is the propeller disc area, is a correction factor that can be adjusted dynamically over time. is the standard temperature, which is 288.12K, or 15°C. is the horizontal flight distance, is the horizontal velocity component, is the windward area; is the zero drag lift coefficient; is the induced drag factor, , Indicates the air density at the current path point height The air density at the next path point The average value of the air density at the current path point height The air density at the next path point It can be calculated based on the following calculation model: .

[0034] Energy consumption during the above takeoff phase The calculation model is based on the energy consumption modeling of the UAV during the process of taking off from the ground to the flight altitude. The energy required for the UAV to climb from the ground to the target altitude is converted into an integral expression. The various actual influencing factors such as the change of air density with altitude, propulsion system efficiency, flight speed, etc. are considered. The temperature change term is used to calculate the energy consumption of the UAV. Used to simulate the standard atmospheric temperature gradient, indicating the impact of the environment on air density and lift, efficiency parameter The loss of the UAV propulsion system is corrected, and the instantaneous power consumption of the entire process is calculated by integration to obtain the total energy consumption estimation result of the take-off phase, so as to achieve a more accurate estimation of the energy demand of the UAV during the vertical take-off phase and improve the authenticity and adaptability of the energy consumption prediction.

[0035] Energy consumption during the landing phase The calculation model uses integration to represent the instantaneous power accumulation of the UAV during the process of descending from the target flight altitude to the ground. The temperature change term is the same as the take-off phase, which is used to represent the temperature change as the altitude decreases over time. Compared with the traditional static estimation method, this model is more in line with the actual aerodynamic characteristics and can be used to optimize landing posture and path selection, thereby reducing unnecessary energy consumption and system load, and improving system energy-saving efficiency and flight safety.

[0036] In the above embodiment, if the ground temperature distribution and wind speed model are known to be relatively complex, a piecewise temperature function may be introduced or fitted based on measured data to replace the standard temperature gradient term.

[0037] In the above embodiment, the propeller operating efficiency, the motor operating efficiency, and the electronic speed regulator operating efficiency can be obtained through actual flight tests or characteristic curves provided by the manufacturer, or can be set as typical constants for approximate estimation when accurate values are not available.

[0038] Energy consumption during the above-mentioned horizontal flight phase The computational model divides horizontal flight energy consumption into two components: air resistance power consumption, which is the energy consumed per unit time by the drone to counteract air resistance in the forward direction; and lift power, which is the power consumed to maintain lift against gravity. These two components are combined to produce the energy consumption during horizontal flight, which is then corrected using the propulsion system efficiency as the denominator. This fully simulates the actual energy requirements of the drone during complex flight, helping to prioritize and optimize the energy efficiency of different horizontal paths. Compared to traditional, simplified models that only estimate drag energy consumption, this implementation is more practical in engineering and is particularly suitable for path planning for long-distance, variable-altitude missions, enhancing the system's ability to refine energy budgets.

[0039] In the above embodiment, the windward area , zero drag lift coefficient , Induced drag factor When actual measurements are not possible, the parameters can be estimated based on the UAV platform combined with the model and Fengdong measurement standard parameters, or approximate constants can be used in combination with the flight scenario to fine-tune the configuration.

[0040] In the preferred embodiment of this application, is 0.8; is 0.8; is 0.8; is 0.94; 0.75 square meters; is 0.13; 15 kg; is 1.313 square meters; is 0.015; is 15 m / s, based on The horizontal velocity component, the vertical downward velocity component, and the vertical upward velocity component are calculated.

[0041] Based on the above calculation model, this application establishes a refined energy consumption model based on the relationship between aerodynamics and energy conversion based on the horizontal flight stage, take-off stage, and landing stage, accurately quantifies the actual energy consumption of the path from each flight stage, takes into account the energy required to overcome gravity in vertical movement and the influence of air resistance in horizontal flight, and combines the working efficiency of key components such as propellers, motors, and electronic speed regulators to improve the authenticity and engineering applicability of energy consumption calculations.

[0042] In some embodiments, the waypoint acquisition step S21 includes: The three-dimensional flight environment model is projected into a two-dimensional plane obstacle map, and the two-dimensional reference path points from the starting point S to the end point T are generated using the improved ellipse tangent map method.

[0043] In some embodiments, the waypoint acquisition step S21 further includes: In the path initialization step S201, a path node set, a candidate node queue, and a collision obstacle set are initialized to empty sets. The path node set is denoted as Path and is used to record valid nodes in the current path. The candidate node queue is denoted as C and is used to temporarily store path nodes to be expanded. The collision obstacle set is denoted as B and is used to record obstacles that collide with the obstacle during the path search. In the bidirectional path search step S202, the starting point S is added to the path node set Path, and the end point T is added to the candidate node queue C as the preset end point to execute the path collision detection step S4, generate a first candidate path set from the starting point S to the end point T, and add the end point T to the path node set Path, add the starting point S to the candidate node queue C as the preset end point to execute the path collision detection step, iteratively generate a second candidate path set from the end point T to the starting point S, establish a bidirectional path topology map based on the first candidate path set and the second candidate path set, record the collision-free path nodes and their connection relationships; based on this, use the bidirectional search strategy to perform path planning, take the starting point S and the end point T as target points, and perform the path collision detection step at the same time to find the obstacle avoidance candidate paths from the starting point S to the end point T and from the end point T to the starting point S.

[0044] In the optimal path screening step S203, the first candidate path set and the second candidate path set are combined, and the total length of each path is calculated using a path cost matrix. Each element in the path cost matrix represents the length of a collision-free path segment from one node to another node. If it is based on weighted length, if there is an obstacle, it is assigned an infinite value. The Dijkstra algorithm is used to select the collision-free path with the minimum total path length in the bidirectional path topology graph, and the two-dimensional reference path point of the global shortest path is output; Among them, during the collision detection process, the tangent graph method is used to generate candidate temporary paths, namely tangent paths, and a temporary path from the candidate temporary paths is selected for path generation based on preset heuristic rules to ensure that the path is reasonable in detouring between obstacles and has high planning efficiency, avoiding falling into local optimality.

[0045] Based on the above steps, a two-dimensional path search framework between the starting point and the end point is constructed using the elliptical tangent graph method, and a bidirectional search strategy is introduced to improve the path convergence efficiency, effectively reducing the redundant search range and significantly improving the search efficiency and performance. Initializing the empty path set and candidate set helps to accurately track the path evolution process. By compressing the three-dimensional terrain into a two-dimensional graph and introducing a topological graph construction method, the obstacle boundary information can be converted into a path connectivity relationship. The Dijkstra algorithm is used in conjunction with the cost matrix, so that the path generation not only considers obstacle avoidance, but also has the ability to optimize the path. The generation of candidate paths is achieved through the "tangent strategy + heuristic selection", making the path bending position more controllable and continuous.

[0046] In the above embodiment, the path cost matrix can be expanded to a weighted cost function including composite indicators such as flight energy consumption and corner penalty.

[0047] In some embodiments, the path collision detection step S4 includes: obtaining an end node O from the path node set Path, obtaining an end node from the candidate node queue C as a target node D, connecting the end node O and the target node D to obtain a path segment OD for collision detection, and if a collision is detected, replanning the path until the path reaches a predetermined end point without encountering any obstacles, thereby generating a candidate path. During reverse iteration, the predetermined end point represents the starting point S.

[0048] When the path collision detection step S4 is used to perform forward and reverse iterations, the path finding quality in a complex obstacle-intensive environment can be improved.

[0049] The path collision detection step S4 further includes: S401: Obtain the end node O in the path node set Path, express it as O=Path[end], and use it as the starting node of the current extended path; S402: Perform collision detection on the path segment OD. If it does not intersect with an obstacle, move the target node D from the candidate node queue C to the path node set Path. Otherwise, execute step S403 to correct the path. S403: Determine the first collision obstacle Obs, obtain the first tangent and the second tangent between the end node O and the collision obstacle Obs, and obtain the third tangent and the fourth tangent between the end point T and the collision obstacle Obs. Based on the intersection point N1 of the first and third tangents and the intersection point N1′ of the second and fourth tangents, two candidate temporary paths are generated, represented as path Path1: S→N1→T and path Path2: S→N1′→T. S404: Determine an optimal intersection point N* between the intersection point N1 and the intersection point N1′ based on a preset heuristic rule as a relay point of the current candidate path; S405: Connect the optimal intersection point N* and the end node O for collision detection. If there is no intersection with the obstacle, that is, no collision, then the optimal intersection point N* is added to the path node set Path, and the optimal intersection point N* is used as the new end node O for loop iteration. Otherwise, the optimal intersection point N* is added to the candidate node queue C, and the collision obstacle Obs is added to the collision obstacle set B. S406: If the candidate node queue C is not empty, obtain the target node D′ in the candidate node queue C, repeat the above steps S402 to S405, and connect the end node O and the target node D′ to perform collision detection.

[0050] Based on the above steps, the embodiment of the present application uses two-point tangent intersection to construct candidate obstacle-avoiding paths, uses intersections to guide the generation of alternative paths, and dynamically selects the optimal path through a heuristic selection mechanism. When there are unprocessed path points in the candidate node queue, the above steps S402 to S405 can be repeated until a complete path to the end point without collision is successfully generated.

[0051] In some embodiments, the preset heuristic rules are configured to include: First condition: select the subpath with the least collision with the starting point S tangent; Second condition: select the subpath that has less collision with the end point T tangent line; The third condition: select the subpath with shorter length; Based on the above preset heuristic rules, collision detection is performed on the S→N1 or S→N1' segment of each path based on the first condition, the number of intersections with obstacles is recorded, and the number of collision obstacles is counted for comparison to ensure that the initial segment of the path avoids complex obstacle areas as much as possible, reducing the risk of initial path failure; If the number of collision obstacles in the first condition is the same, collision detection is performed on the N1→T or N1'→T segment based on the second condition, the number of intersections with obstacles is recorded, and the number of collision obstacles is counted for comparison to ensure that the obstacle interference is minimized when the end of the path approaches the target; If the number of collision obstacles in the second condition is also the same, the shortest path is pursued while meeting the obstacle avoidance requirement, thereby optimizing flight time and energy consumption.

[0052] Based on the above cyclic collision detection process, the optimal path in two-dimensional space is solved according to the principle of tangent intersection.

[0053] In some embodiments, in the flight environment modeling step S1, the top view outer contour of the building is modeled as an ellipse to obtain an obstacle model to represent the restricted area for drone flight. The obstacle model is represented as: , In the above formula, and Respectively represent the semi-major axis and semi-minor axis of the ellipse; is the inclination angle between the semi-major axis and the positive direction of the horizontal axis in the two-dimensional obstacle map; Indicates the safe distance between the drone and obstacles; Indicates the center coordinates of the obstacle; Indicates the height of the obstacle.

[0054] Based on the obstacle model and the actual geometric boundary characteristics of the obstacle in the flight area, the modeling efficiency and path algorithm compatibility are improved by approximating it to an ellipse. In three-dimensional space, the obstacle area is a closed elliptical cylinder, which restricts the drone's flight path, simplifies the obstacle representation in the flight environment, and improves modeling speed and algorithm parsing efficiency. This modeling method is highly parameterized and can achieve approximate representation of the diverse shapes of buildings. By configuring safety distances, the no-fly boundary can be dynamically adjusted to implement flight safety strategies that adapt to different mission risk levels. Compared to polygonal or voxel modeling methods, this solution has low computational overhead, continuous representation, and facilitates geometric operations and shortest path searches in path planning algorithms.

[0055] In the above embodiment, the ellipse parameters 、 It can be obtained by fitting the building's circumscribed rectangle or by extracting it based on actual building geometry surveying data. If the building structure is rectangular or regular polygonal, an enveloping ellipse can be used to approximate the boundary. For multi-story buildings with drastic height changes, Replaced with a multi-segment height function to construct a layered ellipse model. Safety distance It can be set as dynamic parameters related to factors such as flight speed, wind speed, and autonomous control accuracy to achieve differentiated flight safety strategies.

[0056] The two-dimensional obstacle map in the embodiment of this application takes three typical obstacle distributions as examples: Figure 4 The figure shows a two-dimensional obstacle map of low-density obstacle distribution. The obstacles in the figure are circular and are denoted as C1 distribution. Figure 5 The figure shows another two-dimensional obstacle map with low-density obstacle distribution, which includes elliptical obstacles and circular obstacles and is denoted as C2 distribution. Figure 6 The figure shows a two-dimensional obstacle map with high-density obstacle distribution, including elliptical obstacles and circular obstacles, which is denoted as C3 distribution. Figures 4 to 6 In the figure, the circular point is used to indicate the starting point S, and the triangular point is used to indicate the end point T.

[0057] Considering that most buildings are irregular polygonal shapes, the sharp corners of polygons may cause path planning to generate a jagged path. When determining whether the path intersects with an obstacle, it is necessary to process complex edge and vertex intersection detection. Based on the above-mentioned obstacle model, the embodiment of the present application approximates the building into an ellipse, so that the radius of the ellipse can include the actual outer contour and safety distance of the building, avoiding the path from being close to obstacles and improving fault tolerance. The continuous curvature characteristics of the ellipse can naturally guide the algorithm to generate a smoother path, simplify calculations, improve efficiency and ensure the smoothness and safety of the path.

[0058] In some embodiments, the three-dimensional path smoothing step S3 uses a cubic B-spline curve for smoothing. Specifically: Express the waypoints of the flight path as: , the basis function of the cubic B-spline curve is expressed as the following calculation model: , but, and The cubic B-spline curve between is expressed as: .

[0059] The technical effects of the embodiments of the present application are described and illustrated below through preferred embodiments.

[0060] The preferred embodiment of this application is based on Figures 4 to 6 Obstacle distribution is obtained by randomly setting the obstacle heights in C1 distribution, C2 distribution, and C3 distribution, and the obstacle heights are set within the low-altitude feasible area (i.e., 20-120 m). The path planning algorithm (denoted as ECTG) of the embodiment of the present application is compared and analyzed with the A* algorithm, RRT (Rapidly-exploring Random Tree) algorithm, PSO (Particle Swarm Optimization) algorithm, and VFH (Vector Field Histogram) algorithm under these three 3D obstacle distributions, and the following are generated: Figures 7 to 9 The path planning comparison effect diagram shown is as follows. Figure 7 This is a schematic diagram of the path planning comparison effect under C1 distribution. Figure 8 This is a schematic diagram of the path planning comparison effect under C2 distribution. Figure 9 This is a comparison diagram of path planning under C2 distribution. Figures 10 to 12 They are Figures 7 to 9 Schematic diagram of the overhead effect.

[0061] By analyzing the performance of the evaluation algorithm under three evaluation dimensions: path length, number of turns, and energy consumption, the data summary is shown in Table 1.

[0062] Table 1 - Data Summary

[0063] Based on the above analysis, it can be seen that the path planning method of this application takes into account the path length and the tangent path generated when avoiding obstacles, which improves the quality of path planning. In the three obstacle distribution scenarios of C1, C2, and C3, it can generate the shortest path with the least number of turns, and the more complex the environment, the more advantageous it is. Compared with this application, the RRT algorithm generates path points based on a random sampling method, and does not set a mechanism to improve the path quality. It cannot guarantee the optimality of the path and is prone to causing the drone to frequently turn and detour. The A* algorithm, as a grid-based deterministic search algorithm, can generate a feasible path from the starting point to the end point, but the quality of the path it generates is not excellent; the VFH algorithm is a local path planning algorithm with a limited search range. It focuses on avoiding visible obstacles in front and ignores the overall path length.

[0064] Compared with the above algorithms, the embodiments of the present application perform excellently in key indicators such as path length, number of turns, and energy consumption. Among them, compared with the A* algorithm, RRT algorithm, PSO algorithm, and VFH algorithm, the average path length is reduced by 8.7%, 18.7%, 13.4%, and 4.1%, respectively; the average number of turns is reduced by 68.8%, 82.1%, 82.8%, and 75.0%, respectively; and the average energy consumption is reduced by 51.6%, 34.0%, 59.1%, and 55.3%, respectively. This shows that the present application can more efficiently find a short path close to the optimal one, and the generated path is smoother, which is conducive to stable flight of the UAV and reduces the control burden. In particular, it performs outstandingly in terms of energy consumption, helps to extend the UAV's flight time, and also provides strong support for the optimization of battery management strategies in actual missions. As the complexity of the environment increases, its advantages become more significant.

[0065] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0066] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0067] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A UAV path planning method based on ellipse tangent graph method, characterized in that: include: The flight environment modeling step involves determining the flight area, obtaining the geographic location information, size and structure information of buildings within the flight area, constructing an elliptical obstacle model using the buildings as obstacles, and constructing a three-dimensional flight environment model based on the elevation information of the flight area and the distribution of the obstacle model. A three-dimensional path acquisition step is to connect the path ST from the starting point S to the end point T, and perform collision detection on the path ST based on the three-dimensional flight environment model. If no collision occurs, the drone is controlled to fly along the path ST; If a collision occurs, the path point acquisition step is executed to generate a two-dimensional reference path point and determine whether the height of the starting point S and the end point T are the same; If they are the same, a flight path to bypass the obstacle is generated based on the bypass strategy or the climb strategy; If they are different, an alternative path is generated and the flight path with the lowest consumption among the alternative paths is output based on a flight energy consumption model, wherein n path points between the starting point S and the end point T are obtained based on the two-dimensional reference path point, and each path point is configured with a detour strategy and a climb strategy to generate Alternative paths are generated, and the UAV flight action is decomposed into take-off phase, landing phase, and horizontal flight phase according to the alternative paths. Based on the energy consumption of the take-off phase, Energy consumption during landing phase , energy consumption during horizontal flight The flight energy consumption model is constructed, and the flight path with the lowest energy consumption among the alternative paths is output using the flight energy consumption model. The flight energy consumption model is represented by the following calculation model: ； The three-dimensional path smoothing step smoothes the path points of the flight path output by the three-dimensional path acquisition step to obtain a flight trajectory for controlling the UAV.

2. The UAV path planning method based on the ellipse tangent graph method according to claim 1 is characterized in that: Energy consumption during takeoff Calculated based on the following calculation model: ， in, is the ground height, is the flight altitude of the UAV, represents the vertical upward velocity component, , , Respectively represent the propeller working efficiency, motor working efficiency and electronic speed regulator working efficiency, is the total mass of the UAV, is the acceleration due to gravity, , Indicates the air density, the value is 1.225 ; is the flight time, is the temperature gradient below the top of the troposphere, with a value of -0.0065 , R is the gas constant in air, which is 287.05287 , is the propeller disc area, is the correction factor, is the standard temperature, which is 288.12K.

3. The UAV path planning method based on the ellipse tangent graph method according to claim 2 is characterized in that: Energy consumption during landing phase Calculated based on the following calculation model: ， in, They represent the vertical downward velocity components respectively.

4. The UAV path planning method based on the ellipse tangent graph method according to claim 3 is characterized in that: Energy consumption during horizontal flight Calculated based on the following calculation model: ， in, is the horizontal flight distance, is the horizontal velocity component, is the windward area; is the zero drag lift coefficient; is the induced drag factor, , Indicates the air density at the current path point height The air density at the next path point The average value of .

5. The UAV path planning method based on the ellipse tangent graph method according to any one of claims 1 to 4, characterized in that: The path point acquisition step includes: The three-dimensional flight environment model is projected into a two-dimensional plane obstacle map, and the two-dimensional reference path points from the starting point S to the end point T are generated using the improved ellipse tangent map method.

6. The UAV path planning method based on the ellipse tangent graph method according to claim 5 is characterized in that: The path point acquisition step further comprises: Path initialization step: Initialize a path node set, candidate node queue, and collision obstacle set to empty sets. The path node set is represented as Path, the candidate node queue is represented as C, and the collision obstacle set is represented as B. In a bidirectional path search step, the starting point S is added to the path node set Path, and the end point T is added to the candidate node queue C as the preset end point. A path collision detection step is performed to generate a first candidate path set from the starting point S to the end point T. The end point T is then added to the path node set Path, and the starting point S is added to the candidate node queue C as the preset end point. A path collision detection step is performed to iteratively generate a second candidate path set from the end point T to the starting point S. A bidirectional path topology graph is established based on the first and second candidate path sets, and collision-free path nodes and their connection relationships are recorded. an optimal path screening step, combining the first candidate path set and the second candidate path set, calculating the total length of each path using a path cost matrix, selecting a collision-free path with the minimum total length in the bidirectional path topology graph using the Dijkstra algorithm, and outputting a two-dimensional reference path point of the global shortest path; In the collision detection process, a tangent graph method is used to generate candidate temporary paths, and a temporary path among the candidate temporary paths is selected based on a preset heuristic rule to perform path generation.

7. The UAV path planning method based on the ellipse tangent graph method according to claim 6, characterized in that: The path collision detection step comprises: Get the end node O in the path node set Path, get the end node in the candidate node queue C as the target node D, connect the end node O and the target node D to obtain the path segment OD for collision detection. If a collision is detected, replan the path until the path reaches the preset end point without obstacle collision, and generate a candidate path.

8. The UAV path planning method based on the ellipse tangent graph method according to claim 7, characterized in that: The path collision detection step further comprises: Obtain the end node O in the path node set Path; Perform collision detection on the path segment OD. If it does not intersect with the obstacle, move the target node D from the candidate node queue C to the path node set Path. Otherwise, Determine the first collision obstacle Obs, obtain the first and second tangents between the end node O and the collision obstacle Obs, and obtain the third and fourth tangents between the end point T and the collision obstacle Obs. Generate two candidate temporary paths based on the intersection point N1 of the first and third tangents and the intersection point N1′ of the second and fourth tangents. Determine an optimal intersection point N* between the intersection point N1 and the intersection point N1′ based on a preset heuristic rule; Connect the optimal intersection point N* and the end node O for collision detection. If there is no intersection with the obstacle, add the optimal intersection point N* to the path node set Path and use the optimal intersection point N* as the new end node O for loop iteration. Otherwise, add the optimal intersection point N* to the candidate node queue C and add the collision obstacle Obs to the collision obstacle set B. If the candidate node queue C is not empty, obtain the target node D′ in the candidate node queue C, repeat the above steps, and connect the end node O and the target node D′ for collision detection.

9. The UAV path planning method based on the ellipse tangent graph method according to claim 5, characterized in that: In the flight environment modeling step, the top view outer contour of the building is modeled as an ellipse to obtain an obstacle model, which is expressed as: ， In the above formula, and Respectively represent the semi-major axis and semi-minor axis of the ellipse; is the inclination angle between the semi-major axis and the positive direction of the horizontal axis in the two-dimensional obstacle map; Indicates the safe distance between the drone and obstacles; Indicates the center coordinates of the obstacle; Indicates the height of the obstacle.

10. The UAV path planning method based on the ellipse tangent graph method according to any one of claims 1 to 4, characterized in that: The three-dimensional path smoothing step uses a cubic B-spline curve for smoothing.

Images