Unmanned aerial vehicle path planning and obstacle avoidance method based on black-winged kite optimization algorithm
By generating a globally optimal path and performing real-time local obstacle avoidance using the Blackwing Kite optimization algorithm, the problems of convergence speed and obstacle modeling complexity in UAV path planning are solved, achieving fast and safe path planning and obstacle avoidance, which is applicable to a variety of scenarios.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENYANG AEROSPACE UNIVERSITY
- Filing Date
- 2026-01-27
- Publication Date
- 2026-06-05
AI Technical Summary
Existing UAV path planning algorithms suffer from slow convergence speed and are prone to getting stuck in local optima. Furthermore, obstacle modeling is complex and lacks versatility, resulting in delayed obstacle avoidance response and impacting flight safety and mission efficiency.
The Black-winged Kite Optimization Algorithm is used to generate the globally optimal path. By combining attack behavior and migration behavior with the Cauchy mutation strategy, sudden obstacles are detected in real time and local optimal paths are generated. Obstacles are modeled as cylinders to achieve fast convergence and general modeling.
It improves the convergence speed and mission efficiency of UAV path planning, reduces collision avoidance risks, adapts to various scenarios, and has a short response time, ensuring safe flight.
Smart Images
Figure CN122151877A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of autonomous flight technology for unmanned aerial vehicles (UAVs), specifically to a UAV path planning and obstacle avoidance method based on the Blackwing Kite optimization algorithm, applicable to various UAV application scenarios such as logistics transportation, search and rescue, and precision agriculture. Background Technology
[0002] One of the core technologies for autonomous flight of unmanned aerial vehicles (UAVs) is path planning and obstacle avoidance. This requires planning the optimal path (such as the shortest distance and lowest energy consumption) in complex environments and avoiding sudden obstacles in real time. Existing path planning optimization algorithms (such as A* algorithm and particle swarm optimization algorithm) suffer from slow convergence speed and a tendency to get trapped in local optima. Furthermore, the diverse types of obstacles (such as buildings, trees, and other aircraft) and the complexity and poor versatility of modeling methods lead to delayed obstacle avoidance response, affecting flight safety and mission efficiency.
[0003] In recent years, optimization algorithms based on biological behavior have attracted attention due to their efficient search capabilities. However, existing algorithms still have shortcomings in balancing global optimality and real-time obstacle avoidance in UAV path planning. Therefore, there is an urgent need for a UAV path planning and obstacle avoidance scheme that combines fast convergence, general modeling, and real-time obstacle avoidance capabilities. Summary of the Invention
[0004] In view of this, the present invention provides a UAV path planning and obstacle avoidance method based on the Black-winged Kite optimization algorithm to solve the problems existing in traditional UAV path planning and obstacle avoidance methods.
[0005] This invention provides a method for UAV path planning and obstacle avoidance based on the Black-winged Kite optimization algorithm, comprising:
[0006] S1: Generate a 3D map based on the UAV's flight area size, maximum flight altitude, origin and destination, and known obstacle information. For each known obstacle, model it as a cylinder and mark all points inside and on the surface of the cylinder as impassable points.
[0007] S2: Import the 3D map and origin-end point information into the UAV. The UAV uses the Blackwing Kite optimization algorithm to generate the globally optimal path. The Blackwing Kite optimization algorithm includes attack behavior, migration behavior and Cauchy mutation strategy.
[0008] S3: The UAV flies along the globally optimal path and detects sudden obstacles in real time. If an obstacle is detected, the UAV uses the Blackwing Kite optimization algorithm to generate a locally optimal path and flies along the locally optimal path to achieve local obstacle avoidance until it reaches the destination.
[0009] Preferably, in S1, the method for generating the 3D map is as follows:
[0010] S11: Obtain the length l, width w, maximum flight altitude d of the drone, coordinates of the start and end points of the flight area, and the center position c, height h, and longest distance R_obs from the center to the edge of the known obstacles;
[0011] S12: Initialize a 3D array of l×w×d dimensions as an empty map, and mark each point in the array as a passable point;
[0012] S13: For each known obstacle, model it as a cylinder: take the center position c of the obstacle as the origin of the bottom surface, the longest distance from the center to the edge R_obs as the radius of the bottom surface, the height of the obstacle h as the height of the cylinder, and mark all points inside and on the surface of the cylinder as traversable points in the three-dimensional array.
[0013] Further optimization, in S2, the specific steps for generating the globally optimal path using the Black-winged Kite optimization algorithm are as follows:
[0014] S21: Initialize core parameters, wherein the core parameters include attack behavior step size factor n, migration behavior constant m, attack behavior threshold p, number of path points dim, population size pop, maximum number of iterations T, and mutation scaling factor scale_factor.
[0015] S22: Initialize a population array of dimension pop×dim, corresponding to pop paths, wherein the coordinates of the path points in each path are within the range of the three-dimensional map and do not fall into the obstacle area;
[0016] S23: Calculate the fitness value of each path and select the path with the best fitness as the leader;
[0017] S24: Iterative execution of attack behavior: Generate random numbers in the interval [0,1]. ,when When p < p, perform a hovering attack and update the pathpoint coordinates using the following formula:
[0018] ;
[0019] when When p ≥ p, execute a dive attack and update the path point coordinates using the following formula:
[0020] ;
[0021] in, This represents the current iteration number. Generate independent random numbers in the interval [0,1]. For the attack behavior step size factor, and These are the coordinates of the path points before and after the update;
[0022] S25: Perform migration behavior: Calculate the fitness value Fi of the current population and the fitness value Fri of the random population. If Fri < Fi, the leader relinquishes leadership, updates the path point coordinates, and the update formula is:
[0023] ;
[0024] If Fri ≥ Fi, the leader continues to guide the group, updating the waypoint coordinates using the following formula:
[0025] ;
[0026] in, For Cauchy random numbers, Let Cauchy be the mathematical independent variable. , Generate independent random numbers in the interval [0,1]. This is the scaling factor for the variation. Indicates the current optimal path;
[0027] S26: Iterate until the maximum number of iterations T is reached, and output the leader path as the globally optimal path.
[0028] Further optimizations include: the attack behavior threshold p ranges from 0.3 to 0.5; the number of path points dim ranges from 5 to 20; the population size pop ranges from 30 to 50; the maximum number of iterations T ranges from 100 to 300; and the mutation scaling factor scale_factor ranges from 0.01 to 0.1.
[0029] Further optimization involves using a fitness function in S23 that considers path length, path smoothness, and obstacle avoidance safety indicators.
[0030] Further optimization, in S24, the attack behavior step size factor In the formula, This represents the current iteration number. This represents the maximum number of iterations.
[0031] Further optimization involves using a visual camera or lidar in S3 to detect sudden obstacles, with a detection range of 5 to 20 meters in front of the drone's flight direction.
[0032] Further optimization, in S3, the local optimal path planning method after detecting a sudden obstacle is as follows:
[0033] S31: Obtain the center position new_c, height new_h, and longest distance from the center to the edge new_R_obs of the sudden obstacle;
[0034] S32: According to the modeling rules of S1, model the sudden obstacle as a cylinder and update the 3D map;
[0035] S33: Using the current position of the drone as the new starting point and the original endpoint as the endpoint, repeat the Blackwing Kite optimization algorithm of S2 to replan the local optimal path.
[0036] The UAV path planning and obstacle avoidance method based on the Black-winged Kite Optimization Algorithm provided by this invention accelerates the convergence speed of UAV path planning and local obstacle avoidance by utilizing the Black-winged Kite Optimization Algorithm, improves task efficiency and reduces collision risk. By uniformly treating obstacles as cylinders and setting them as impassable points, the modeling difficulty and complexity are reduced. Moreover, this obstacle determination method is universal and can be applied to a variety of scenarios. Attached Figure Description
[0037] Figure 1 This is a flowchart of the UAV path planning and obstacle avoidance method based on the Black-winged Kite optimization algorithm provided by the present invention;
[0038] Figure 2 A rendering of the obstacle model. Detailed Implementation
[0039] The present invention will be further described below with reference to specific embodiments.
[0040] In order to solve the problems existing in the current technology, such as Figure 1 As shown, this invention provides a method for UAV path planning and obstacle avoidance based on the Black-winged Kite optimization algorithm, including the following steps:
[0041] S1: Generate a 3D map based on the UAV's flight area size, maximum flight altitude, start and end points, and known obstacle information (center position, altitude, and longest distance from the center to the edge). The map can be represented as a raster image.
[0042] The specific steps are as follows:
[0043] S11: Obtain the dimensions of the flight area (length l, width w), the maximum flight altitude d of the drone, the coordinates of the start and end points, and the information of known obstacles (center position c, height h, and the longest distance from the center to the edge R_obs).
[0044] S12: Initialize a 3D array (city_map) with l×w×d dimensions as an empty map, with array elements initially set to 0 (representing passable points);
[0045] S13: For each known obstacle, model it as a cylinder (e.g., ...). Figure 2As shown): with the center position c of the obstacle as the origin of the bottom surface, the longest distance R_obs from the center to the edge as the radius of the bottom surface, and the height h of the obstacle as the height of the cylinder, all points inside and on the surface of the cylinder are marked as 1 in the three-dimensional array (representing points that cannot be passed through).
[0046] S14: Explicitly mark the start and end points in the three-dimensional array (mark value is 2) to ensure the uniqueness of the start and end points of the path planning;
[0047] This modeling method transforms all types of obstacles into standardized geometric shapes, reducing the processing complexity of different types of obstacles. It is applicable to various scenarios such as fixed ground obstacles and suspended aerial obstacles, and has strong versatility.
[0048] S2: Import the 3D map and origin-end point information into the UAV. The UAV uses the Blackwing Kite optimization algorithm to generate the globally optimal path. The Blackwing Kite optimization algorithm includes attack behavior, migration behavior and Cauchy mutation strategy.
[0049] The core logic of the Black-winged Kite Optimization Algorithm for generating globally optimal paths simulates the attack behavior (hovering attack, dive attack) and migration behavior of the Black-winged Kite, and combines it with the Cauchy mutation strategy to escape local optima. The specific steps are as follows:
[0050] S21: Initialize core parameters, wherein the core parameters include attack behavior step size factor n, migration behavior constant m, attack behavior threshold p (0.3~0.5), number of path points dim (5~20), population size pop (30~50), maximum number of iterations T (100~300), and mutation scaling factor scale_factor (0.01~0.1).
[0051] S22: Initialize a population array pop_pos of dimension pop×dim, corresponding to pop paths. Each element in the population array corresponds to a path point coordinate. Each path point coordinate is constrained within the range of the three-dimensional map and does not fall into the obstacle area marked with a value of 1.
[0052] S23: Calculate the fitness value of each path and select the path with the best fitness as the leader (Xbest).
[0053] The fitness function can consider metrics such as path length, path smoothness, and obstacle avoidance safety, balancing efficiency and safety in path planning. For example, the fitness function Fit = α × L + β × S + γ × O, where:
[0054] L is the path length (the sum of the Euclidean distances between all path points between the start and end points);
[0055] S represents the path smoothness (the mean cosine of the angle between the lines connecting adjacent path points).
[0056] O represents the obstacle penalty (when the minimum distance between the path and the obstacle is less than the safety threshold, a penalty value is added inversely proportional to the distance).
[0057] α, β, and γ are weight coefficients, and α+β+γ=1, such as: α=0.5, β=0.3, γ=0.2;
[0058] Then, the path with the smallest fitness value is selected as the leader (Xbest).
[0059] S24: Iterative execution of attack behavior: Generate random numbers in the interval [0,1]. ,when When p < p, perform a hovering attack and update the pathpoint coordinates using the following formula: Simulates the behavior of a black-winged kite hovering and searching for prey to achieve precise local searches; when When p ≥ p, execute a dive attack and update the path point coordinates using the following formula: ,in, This represents the current iteration number. For independent random numbers in the interval [0,1] (and) (Unrelated), simulating the rapid movement of a black-winged kite swooping down to hunt, improving global search efficiency; optimal, attack behavior step size factor. , The current iteration number is used, and it gradually decreases as the iteration number increases, achieving "wide-area search in the early stage and fine convergence in the later stage";
[0060] S25: Enforcing migration behavior:
[0061] Calculate the fitness value (Fi) of the current population and the fitness value (Fri) of the random population:
[0062] If Fri < Fi (the current path is poor), the leader abandons leadership and joins the migrating group, updating the path point coordinates using the following formula: ;
[0063] If Fri ≥ Fi (the current path is better), the leader continues to guide the group to the destination, updating the waypoint coordinates using the following formula: ,in, For Cauchy random numbers, The independent variable is a Cauchy distribution (the range of real numbers, with no fixed interval). , Using independent random numbers in the interval [0,1], Cauchy mutation is used to increase population diversity, helping the algorithm escape local optima. This is the scaling factor for the variation. Indicates the current optimal path;
[0064] S26: Iterate until the maximum number of iterations T is reached, and output the leader path (Xbest) as the globally optimal path;
[0065] S3: The UAV flies along the global optimal path and detects sudden obstacles in real time. If an obstacle is detected, the UAV generates a local optimal path using the Blackwing Kite optimization algorithm and flies along the local optimal path to achieve local obstacle avoidance until it reaches the destination.
[0066] During flight, if the drone does not detect any sudden obstacles, it flies according to the path point sequence of the globally optimal path. When a sudden obstacle is detected (by a visual camera or lidar), for example, if the detection range is 5-20 meters in front of the drone's flight direction, and the obstacle enters this range and is located on the current flight path, local obstacle avoidance is triggered, specifically as follows:
[0067] S31: Obtain the center position (new_c), height (new_h), and longest distance from the center to the edge (new_R_obs) of the sudden obstacle.
[0068] S32: According to the modeling rules of S1, model the sudden obstacle as a cylinder and update the 3D map (add the sudden obstacle to the 3D map and update the array markers).
[0069] S33: Using the current position of the drone as the new starting point and the original endpoint as the endpoint, repeat the Blackwing Kite optimization algorithm of S2 to replan the local optimal path (the number of iterations can be adjusted to 50~100 times to improve real-time performance).
[0070] The UAV path planning and obstacle avoidance method based on the Black-winged Kite Optimization Algorithm provided by this invention utilizes the Black-winged Kite Optimization Algorithm to achieve "wide-area search + fine convergence" through attack behaviors (hovering + dive), and combines Cauchy mutation strategy to avoid local optima. Compared with the traditional particle swarm optimization algorithm, the convergence speed is significantly improved. All obstacles are uniformly modeled as cylinders, which reduces the modeling complexity and is suitable for various application scenarios. The response time of sudden obstacle detection and local path replanning is short, ensuring that the UAV can avoid obstacles in time.
[0071] Example 1
[0072] This embodiment takes a precision agriculture scenario as an example. The drone needs to fly in a farmland area (length l=100m, width w=80m, maximum flight height d=50m), with the starting coordinates (0,0,20) and the ending coordinates (100,80,20). The known obstacles are 3 farmland irrigation towers (the center positions are (30,30,15), (60,50,20), and (80,20,18) respectively, with a height of 30m and the longest distance R_obs from the center to the edge is 5m).
[0073] The specific methods for drone path planning and obstacle avoidance are as follows:
[0074] S1: 3D map generation
[0075] Initialize a 3D array with dimensions of 100×80×50. Model the three irrigation towers as cylinders. Mark the points inside and on the surface of the cylinders as 1, the starting point (0,0,20) and the ending point (100,80,20) as 2, and the remaining points as 0.
[0076] S2: Global Path Planning
[0077] Parameter settings: p=0.4, dim=10, pop=40, T=200, scale_factor=0.05. After 200 iterations using the Black-winged Kite optimization algorithm, the globally optimal path was obtained, with a path length of 135.2m and a minimum distance to obstacles ≥8m, meeting the safety requirements.
[0078] Attack execution: Each iteration generates independent r_rand and r_attack. r_rand is used to select hover / dive mode, and r_attack dynamically adjusts the dive intensity, direction switching or policy trigger probability.
[0079] Migration behavior execution: r_migrate is generated independently to calculate the m value, and combined with the Cauchy random numbers generated by x_cauchy to achieve population mutation and avoid local optima;
[0080] S3: Sudden Obstacle Avoidance
[0081] When the drone flew to the position (50,40,20), it detected a sudden obstacle (a flock of birds, centered at (55,45,18), at a height of 25m, with the longest distance from the center to the edge new_R_obs=3m). It immediately updated the map and replanned the local path starting from (50,40,20). After 80 iterations, a new path was obtained, and the drone flew along the new path to the destination. The obstacle avoidance response time was 0.8 seconds, and the path length increased by 5.3m, without affecting the mission execution efficiency.
[0082] Example 2
[0083] This embodiment compares the performance of the algorithm of this invention with that of the traditional particle swarm optimization (PSO) algorithm:
[0084] Convergence speed: The algorithm of this invention reaches the optimal path in 85 iterations, while the PSO algorithm takes 142 iterations, representing a 40.1% improvement in convergence speed;
[0085] Path length: The average path length planned by this invention is 138.5m, while that of the PSO algorithm is 152.3m, making the path shorter;
[0086] Obstacle avoidance success rate: In 5 groups of experiments with sudden obstacles, the obstacle avoidance success rate of the present invention was 100%, while that of the PSO algorithm was 80% (2 groups failed due to convergence lag).
[0087] The above embodiments demonstrate that the UAV path planning and obstacle avoidance method based on the Black-winged Kite optimization algorithm of the present invention is superior to the prior art in terms of convergence speed, path optimization and obstacle avoidance safety. The technical solution has strong repeatability and broad application prospects.
[0088] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.
[0089] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.
Claims
1. A method for unmanned aerial vehicle path planning and obstacle avoidance based on a black-winged kite optimization algorithm, characterized in that, include: S1: Generate a 3D map based on the UAV's flight area size, maximum flight altitude, origin and destination, and known obstacle information. For each known obstacle, model it as a cylinder and mark all points inside and on the surface of the cylinder as impassable points. S2: Import the 3D map and origin-end point information into the UAV. The UAV uses the Blackwing Kite optimization algorithm to generate the globally optimal path. The Blackwing Kite optimization algorithm includes attack behavior, migration behavior and Cauchy mutation strategy. S3: The UAV flies along the globally optimal path and detects sudden obstacles in real time. If an obstacle is detected, the UAV uses the Blackwing Kite optimization algorithm to generate a locally optimal path and flies along the locally optimal path to achieve local obstacle avoidance until it reaches the destination.
2. The method according to claim 1, wherein the method is characterized in that: In S1, the method for generating the 3D map is as follows: S11: Obtain the length l, width w, maximum flight altitude d of the drone, coordinates of the start and end points of the flight area, and the center position c, height h, and longest distance R_obs from the center to the edge of the known obstacles; S12: Initialize a 3D array of l×w×d dimensions as an empty map, and mark each point in the array as a passable point; S13: For each known obstacle, model it as a cylinder: take the center position c of the obstacle as the origin of the bottom surface, the longest distance from the center to the edge R_obs as the radius of the bottom surface, the height of the obstacle h as the height of the cylinder, and mark all points inside and on the surface of the cylinder as traversable points in the three-dimensional array.
3. The UAV path planning and obstacle avoidance method based on the Black-winged Kite optimization algorithm according to claim 1, characterized in that: In S2, the specific steps for generating the globally optimal path using the Black-winged Kite optimization algorithm are as follows: S21: Initialize core parameters, wherein the core parameters include attack behavior step size factor n, migration behavior constant m, attack behavior threshold p, number of path points dim, population size pop, maximum number of iterations T, and mutation scaling factor scale_factor. S22: Initialize a population array of dimension pop×dim, corresponding to pop paths, wherein the coordinates of the path points in each path are within the range of the three-dimensional map and do not fall into the obstacle area; S23: Calculate the fitness value of each path and select the path with the best fitness as the leader; S24: Iterative execution of attack behavior: Generate random numbers in the interval [0,1]. ,when When p < p, perform a hovering attack and update the pathpoint coordinates using the following formula: ； when When p ≥ p, execute a dive attack and update the path point coordinates using the following formula: ； in, This represents the current iteration number. Generate independent random numbers in the interval [0,1]. For the attack behavior step size factor, and These are the coordinates of the path points before and after the update; S25: Perform migration behavior: Calculate the fitness value Fi of the current population and the fitness value Fri of the random population. If Fri < Fi, the leader relinquishes leadership, updates the path point coordinates, and the update formula is: ； If Fri ≥ Fi, the leader continues to guide the group, updating the waypoint coordinates using the following formula: ； in, For Cauchy random numbers, Let Cauchy be the mathematical independent variable. , Generate independent random numbers in the interval [0,1]. This is the scaling factor for the variation. Indicates the current optimal path; S26: Iterate until the maximum number of iterations T is reached, and output the leader path as the globally optimal path.
4. The UAV path planning and obstacle avoidance method based on the Black-winged Kite optimization algorithm according to claim 3, characterized in that: The attack behavior threshold p ranges from 0.3 to 0.5, the number of path points dim ranges from 5 to 20, the population size pop ranges from 30 to 50, the maximum number of iterations T ranges from 100 to 300, and the mutation scaling factor scale_factor ranges from 0.01 to 0.
1.
5. The UAV path planning and obstacle avoidance method based on the Black-winged Kite optimization algorithm according to claim 3, characterized in that: The fitness function used in S23 considers path length, path smoothness, and obstacle avoidance safety indicators.
6. The UAV path planning and obstacle avoidance method based on the Black-winged Kite optimization algorithm according to claim 3, characterized in that: In S24, the attack behavior step size factor In the formula, This represents the current iteration number. This represents the maximum number of iterations.
7. The UAV path planning and obstacle avoidance method based on the Black-winged Kite optimization algorithm according to claim 1, characterized in that: In S3, sudden obstacles are detected by visual cameras or lidar, with a detection range of 5 to 20 meters in front of the drone's flight direction.
8. The UAV path planning and obstacle avoidance method based on the Black-winged Kite optimization algorithm according to claim 1, characterized in that: In S3, the local optimal path planning method after detecting a sudden obstacle is as follows: S31: Obtain the center position new_c, height new_h, and longest distance from the center to the edge new_R_obs of the sudden obstacle; S32: According to the modeling rules of S1, model the sudden obstacle as a cylinder and update the 3D map; S33: Using the current position of the drone as the new starting point and the original endpoint as the endpoint, repeat the Blackwing Kite optimization algorithm of S2 to replan the local optimal path.