A method and system for UAV path planning based on fusion algorithm

Abstract

This invention provides a method and system for UAV path planning based on a fusion algorithm, relating to the field of UAV autonomous navigation and path planning technology. The method includes the following steps: acquiring the starting coordinates, ending coordinates, and obstacle information in the flight environment of a flight mission, and constructing a flight environment spatial model; defining path constraints and optimization objectives, and generating an initial set of feasible paths based on the flight environment spatial model, path constraints, and optimization objectives; performing weighted fusion and guided enhancement on the initial set of feasible paths to obtain a weighted fused path and a guided enhanced path; employing a particle swarm optimization algorithm to construct an initial particle swarm based on the initial feasible path set, the weighted fused path, and the guided enhanced path, and iteratively optimizing the initial particle swarm to obtain a globally optimal particle; and decoding the globally optimal particle to obtain the final flight path. This invention significantly improves the efficiency and quality of UAV path planning in complex static obstacle environments.

Description

Technical Field

[0001] This invention relates to the field of autonomous navigation and path planning technology for unmanned aerial vehicles (UAVs), and in particular to a UAV path planning method and system based on a fusion algorithm. Background Technology

[0002] With the rapid development of drone technology, drones have been widely used in various fields such as logistics and transportation, disaster relief, environmental monitoring, and agricultural plant protection. In these applications, drones typically need to perform autonomous flight missions in complex, unknown, or environments with static / dynamic obstacles. Path planning, as a core component of drone autonomous navigation, aims to plan an optimal or feasible flight trajectory from the starting point to the destination for the drone while meeting safety obstacle avoidance and kinematic constraints. The quality of path planning directly affects the efficiency, flight safety, and energy economy of drone mission execution.

[0003] Currently, UAV path planning methods can be mainly divided into three categories: graph search-based methods (such as A, Dijkstra), random sampling-based methods (such as RRT, RRT-Connect), and intelligent optimization-based methods (such as Particle Swarm Optimization (PSO) and Genetic Algorithm (GA). Each type of method has its own advantages and disadvantages. To balance planning efficiency and path quality, researchers have recently attempted to combine sampling methods with optimization methods to form a hybrid path planning framework. For example, using... The initial path is quickly generated, and then PSO is used to smooth and optimize the path. However, existing hybrid methods still have the following limitations:

[0004] First, most existing methods rely solely on the result of a single RRT run as the initial solution for PSO optimization. Due to the randomness of the RRT algorithm, a single run may yield a severely detoured and poor-quality path, causing subsequent optimization processes to heavily depend on the quality of the initial solution. This can easily lead to getting trapped in local optima or requiring numerous iterations to converge, resulting in low overall planning efficiency. Second, existing methods fail to effectively integrate the effective topological information from multiple sampled paths, resulting in insufficient exploration of the solution space. Furthermore, in complex obstacle environments, path search lacks directional prior knowledge (such as global directional guidance from the start to the end point), leading to significant search blindness, low planning efficiency, and difficulty in generating paths geometrically close to the global optimum. Finally, existing PSO optimizations often perform full-dimensional optimization on all nodes of the entire path, resulting in high computational complexity. Additionally, their parameters are often fixed, making it difficult to adapt to different environmental complexities and optimization stages. This leads to an imbalance between the algorithm's early global exploration capabilities and later local development capabilities, resulting in slow convergence and a tendency for over-optimization or premature convergence.

[0005] How to solve the above-mentioned technical problems is the challenge facing this invention. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a UAV path planning method and system based on a fusion algorithm that significantly improves the efficiency and quality of UAV path planning in complex static obstacle environments.

[0007] The technical solution adopted by this invention to solve its technical problem is as follows: This invention provides a UAV path planning method based on a fusion algorithm, comprising the following steps:

[0008] Obtain the starting coordinates, ending coordinates, and obstacle information in the flight environment of the flight mission, and construct a flight environment spatial model based on the starting coordinates, ending coordinates, and obstacle information in the flight environment.

[0009] Define path constraints and optimization objectives. Based on the flight environment spatial model, path constraints, and optimization objectives, use an improved fast expanding random tree algorithm to generate an initial set of feasible paths. The initial set of feasible paths includes a main path and several secondary paths.

[0010] The initial set of feasible paths is weighted and enhanced to obtain the weighted fused path and the enhanced path.

[0011] The particle swarm optimization algorithm is adopted to construct an initial particle swarm based on the initial feasible path set, weighted fusion path and guided enhancement path, and to iteratively optimize the initial particle swarm to obtain the globally optimal particle.

[0012] The global optimal particle is decoded to obtain the final flight path.

[0013] Preferably, the construction of the flight environment space model includes:

[0014] The system acquires the starting coordinates, ending coordinates, and obstacle information in the flight environment for the flight mission; the obstacle information includes at least the location and size of the obstacles.

[0015] The flight boundary is determined based on the starting point coordinates and the ending point coordinates. An initial obstacle geometric model is constructed based on obstacle information, and the initial obstacle geometric model is expanded to obtain the obstacle geometric model.

[0016] A two-dimensional continuous space is constructed based on the geometric model of the flight boundary and obstacles to obtain the flight environment space model.

[0017] Preferably, the path constraints include obstacle avoidance constraints and path tortuosity constraints, and the optimization objective is to minimize the path length, path tortuosity, and algorithm execution time, as expressed by the following formula:

[0018]

[0019] in, For path length, The path tortuosity. For the algorithm execution time, These are the path length weight, path tortuosity weight, and algorithm execution time weight, respectively.

[0020] Preferably, generating the initial set of feasible paths includes:

[0021] Based on the flight environment spatial model, path constraints, and optimization objectives, an improved fast extended random tree algorithm is executed using a multi-threaded parallel computing method to generate several initial paths;

[0022] Calculate the path length of several initial paths, and filter them according to the path length. The initial path with the shortest path length is selected as the main path, and the initial path with a path length greater than the shortest path length is selected as the secondary path.

[0023] By integrating the main path and the secondary paths, an initial set of feasible paths is obtained.

[0024] Preferably, the improved fast expanding random tree algorithm includes a target-oriented sampling strategy, a parent node selection strategy, and a search radius adjustment strategy;

[0025] The target-oriented sampling strategy is to preset the target bias probability and use a random sampling method to sample target points based on the target bias probability.

[0026] The parent node selection strategy includes evaluating candidate parent nodes within the search radius when expanding to a new node, and selecting the node with the minimum cumulative cost from the starting point to the new node as the parent node; the cumulative cost is calculated using the cumulative Euclidean distance, as shown in the following formula:

[0027]

[0028] in, For the new node Centered on, search radius The set of all nodes within, The distance is Euclidean.

[0029] The search radius adjustment strategy involves adjusting the search radius according to the size of the expanded tree, as expressed by the following formula:

[0030]

[0031] in, As an adjustment constant, The number of nodes in the tree. For spatial dimensions, As a multiple factor, To extend the step size.

[0032] Preferably, the weighted fusion includes:

[0033] The main path and several secondary paths are resampled to the same number of nodes using linear interpolation to obtain the aligned main path and several aligned secondary paths;

[0034] The aligned main path and several aligned secondary paths are weighted and merged using random weighted interpolation to obtain a weighted merged path.

[0035] The guidance enhancements include:

[0036] A straight line is determined based on the starting point coordinates and the ending point coordinates as the sampling line;

[0037] Samples are taken uniformly along the sampling line to obtain several sampling points;

[0038] Based on the obstacle geometry model, collision detection is performed on each sampling point, and sampling points that do not collide are retained as guide points to obtain a set of guide points;

[0039] Define a guide point insertion threshold. For each guide point, if the vertical distance from the guide point to the main path is less than the guide point insertion threshold, then insert the guide point into the main path to obtain the enhanced guide path.

[0040] Preferably, the iterative optimization of the initial particle swarm includes:

[0041] The particle swarm optimization algorithm is used to encode the main path, slave path, weighted fusion path and guided enhancement path as particles to obtain the initial particle swarm.

[0042] Determine the optimal path in the initial particle swarm based on the optimization objective, and calculate the change in turning angle at each intermediate point in the optimal path;

[0043] Define a steering angle change threshold. If the steering angle change value of the intermediate point is greater than the steering angle change threshold, then mark the intermediate point and several adjacent intermediate points as the key optimization area.

[0044] By integrating all key optimization regions, the region to be optimized is obtained.

[0045] An adaptive particle swarm optimization model is constructed, and iterative optimization is performed based on the adaptive particle swarm optimization model and the region to be optimized until the algorithm converges. The adaptive particle swarm optimization model includes a dynamic parameter adjustment strategy, which adaptively adjusts the model parameters according to the iteration progress.

[0046] This invention also provides a UAV path planning system based on a fusion algorithm, comprising:

[0047] The data acquisition and spatial modeling module is used to acquire the starting coordinates, ending coordinates, and obstacle information in the flight environment of the flight mission, and to construct a spatial model of the flight environment based on the starting coordinates, ending coordinates, and obstacle information in the flight environment of the flight mission.

[0048] The initial feasible path generation module is used to define path constraints and optimization objectives. Based on the flight environment spatial model, path constraints, and optimization objectives, it uses an improved fast expanding random tree algorithm to generate an initial feasible path set. The initial feasible path set includes a main path and several secondary paths.

[0049] The weighted fusion and guidance enhancement module is used to perform weighted fusion and guidance enhancement on the initial set of feasible paths to obtain the weighted fusion path and the guidance enhancement path.

[0050] The particle optimization module is used to construct an initial particle swarm based on an initial feasible path set, a weighted fusion path, and a guided enhancement path using the particle swarm optimization algorithm, and to iteratively optimize the initial particle swarm to obtain the globally optimal particle.

[0051] The final path generation module is used to decode the globally optimal particle to obtain the final flight path.

[0052] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described UAV path planning method based on the fusion algorithm.

[0053] The present invention also provides a computer storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described UAV path planning method based on a fusion algorithm.

[0054] The beneficial effects of this invention are as follows: it significantly improves the efficiency and quality of UAV path planning in complex static obstacle environments. It employs an improved RRT-Connect algorithm with multi-threaded parallel execution. This strategy effectively balances the diversity and time overhead of path exploration, overcoming the instability of single-sample randomness. It introduces a multi-path weighted fusion and global guide point enhancement mechanism, creatively utilizing the topological information of multiple initial paths and prior knowledge of the direction from the start to the end point. This avoids excessive reliance on a single path during the optimization process, effectively preventing local optima, and generating paths that are closer to the global optimum and have better geometric features, thus improving the overall quality and rationality of the paths. By identifying key optimization regions for dimensionality reduction and combining them with an adaptive particle swarm optimization model, the algorithm can rapidly explore in the early stages of iteration and refine in the later stages, significantly reducing unnecessary computation and accelerating convergence. Attached Figure Description

[0055] Figure 1 This is a diagram illustrating the method steps of the present invention.

[0056] Figure 2 This is a system module diagram of the present invention.

[0057] Figure 3 This is a schematic diagram of the multi-narrow channel model of Embodiment 3 of the present invention.

[0058] Figure 4 This is a schematic diagram of a cluttered multi-obstacle model according to Embodiment 3 of the present invention.

[0059] Figure 5 This is a schematic diagram of the path results obtained by the method of the present invention in Embodiment 3 of the present invention through 100 independent cycle tests under a multi-narrow channel model.

[0060] Figure 6 This is a schematic diagram of the path results obtained by the RRT-PSO algorithm in Embodiment 3 of the present invention after 100 independent loop tests in a multi-narrow channel model.

[0061] Figure 7 In Embodiment 3 of the present invention A schematic diagram of the path results obtained by the algorithm in 100 independent loop tests under the multi-narrow channel model.

[0062] Figure 8 This is a schematic diagram of the path results obtained by the method of the present invention in 100 independent cyclic tests under a chaotic multi-obstacle model in Embodiment 3 of the present invention.

[0063] Figure 9 In Embodiment 3 of the present invention A schematic diagram of the path results obtained by the algorithm in 100 independent loop trials under a chaotic multi-obstacle model.

[0064] Figure 10 In Embodiment 3 of the present invention A schematic diagram of the path results obtained by the algorithm in 100 independent loop trials under a chaotic multi-obstacle model.

[0065] Figure 11 This is a schematic diagram of the path results obtained by the RRT-PSO algorithm in Embodiment 3 of the present invention after 100 independent loop tests under a chaotic multi-obstacle model.

[0066] Figure 12 This is a diagram of the internal structure of a computer device according to Embodiment 4 of the present invention. Detailed Implementation

[0067] To clearly illustrate the technical features of this solution, the following detailed implementation method will be used to explain the solution.

[0068] Example 1:

[0069] See Figure 1 As shown, this embodiment is a UAV path planning method based on a fusion algorithm, including the following steps:

[0070] S1. Obtain the starting coordinates, ending coordinates, and obstacle information in the flight environment of the flight mission, and construct a flight environment spatial model based on the starting coordinates, ending coordinates, and obstacle information in the flight environment of the flight mission.

[0071] Constructing a flight environment space model includes:

[0072] The system acquires the starting coordinates, ending coordinates, and obstacle information in the flight environment for the flight mission; the obstacle information includes at least the location and size of the obstacles.

[0073] The flight boundary is determined based on the starting point coordinates and the ending point coordinates. An initial obstacle geometric model is constructed based on obstacle information, and the initial obstacle geometric model is expanded to obtain the obstacle geometric model.

[0074] A two-dimensional continuous space is constructed based on the geometric model of the flight boundary and obstacles to obtain the flight environment space model.

[0075] It should be noted that obstacles are represented by a series of convex polygons in two-dimensional space. To simplify collision detection calculations, this embodiment uses rectangles as the basic shape of obstacles. Each rectangular obstacle... By quadruple Define , where represents the coordinates of its lower left corner, width, and height, respectively. The initial set of obstacles can be represented as:

[0076]

[0077] To ensure path safety, a safe distance is added to the obstacles. The geometric model of the obstacle is obtained:

[0078]

[0079] The safe distance is determined based on the physical dimensions of the drone's fuselage, positioning error, and safety margin, and is usually taken as 1.2 to 1.5 times the maximum diagonal length of the drone.

[0080] S2. Define path constraints and optimization objectives. Based on the flight environment spatial model, path constraints, and optimization objectives, use an improved fast expanding random tree algorithm to generate an initial set of feasible paths. The initial set of feasible paths includes a main path and several secondary paths.

[0081] Path constraints include obstacle avoidance constraints and path tortuosity constraints. The optimization objective is to minimize path length, path tortuosity, and algorithm execution time, as expressed in the following formula:

[0082]

[0083] in, For path length, The path tortuosity. For the algorithm execution time, These are the path length weight, path tortuosity weight, and algorithm execution time weight, respectively.

[0084] It should be noted that a route starts from the origin To the finish line path It consists of a series of waypoints:

[0085]

[0086] The path satisfies obstacle avoidance constraints and path tortuosity constraints, where the obstacle avoidance constraint is any continuous line segment on the path. Obstacle geometry model must not be associated with any obstacle. Intersection. The tortuosity constraint is that the path tortuosity must not exceed a preset tortuosity threshold.

[0087] The formula for calculating tortuosity is as follows:

[0088]

[0089] in, For vectors with vector The smaller the angle between the two points, the smoother the path.

[0090] The formula for calculating path length is as follows:

[0091]

[0092] Generating the initial set of feasible paths includes:

[0093] Based on the flight environment spatial model, path constraints, and optimization objectives, an improved fast extended random tree algorithm is executed using a multi-threaded parallel computing method to generate several initial paths;

[0094] Calculate the path length of several initial paths, and filter them according to the path length. The initial path with the shortest path length is selected as the main path, and the initial path with a path length greater than the shortest path length is selected as the secondary path.

[0095] By integrating the main path and the secondary paths, an initial set of feasible paths is obtained.

[0096] The improved fast expanding random tree algorithm includes a target-oriented sampling strategy, a parent node selection strategy, and a search radius adjustment strategy;

[0097] The target-oriented sampling strategy is to preset the target bias probability and then use random sampling to sample target points based on the target bias probability.

[0098] The parent node selection strategy involves evaluating candidate parent nodes within the search radius when expanding to a new node, and selecting the node with the minimum cumulative cost from the starting point to the new node as the parent node; the cumulative cost is calculated using the cumulative Euclidean distance, as shown in the following formula:

[0099]

[0100] in, For the new node Centered on, search radius The set of all nodes within, The distance is Euclidean.

[0101] The search radius adjustment strategy is to adjust the search radius according to the size of the expanded tree, as expressed by the following formula:

[0102]

[0103] in, As an adjustment constant, The number of nodes in the tree. For spatial dimensions, As a multiple factor, To extend the step size.

[0104] S3. Perform weighted fusion and guided enhancement on the initial set of feasible paths to obtain the weighted fused path and the guided enhanced path;

[0105] Weighted fusion includes:

[0106] The main path and several secondary paths are resampled to the same number of nodes using linear interpolation to obtain the aligned main path and several aligned secondary paths;

[0107] The aligned main path and several aligned secondary paths are weighted and merged using random weighted interpolation to obtain a weighted merged path.

[0108] Enhanced guidance includes:

[0109] A straight line is determined based on the starting point coordinates and the ending point coordinates as the sampling line;

[0110] Samples are taken uniformly along the sampling line to obtain several sampling points;

[0111] Based on the obstacle geometry model, collision detection is performed on each sampling point, and sampling points that do not collide are retained as guide points, resulting in a set of guide points:

[0112]

[0113] Define a guide point insertion threshold. For each guide point, if the vertical distance from the guide point to the main path is less than the guide point insertion threshold, then insert the guide point into the main path to obtain the enhanced guide path.

[0114] It should be noted that by using weighted fusion, the effective topological information from multiple initial paths is fully utilized, avoiding excessive reliance on a single path in the optimization process. The guiding point reflects the global shortest direction from the starting point to the ending point, providing important directional prior knowledge for path fusion.

[0115] S4. Using the particle swarm optimization algorithm, an initial particle swarm is constructed based on the initial feasible path set, weighted fusion path, and guided enhancement path, and the initial particle swarm is iteratively optimized to obtain the globally optimal particle.

[0116] Iterative optimization of the initial particle swarm includes:

[0117] The particle swarm optimization algorithm is used to encode the main path, slave path, weighted fusion path and guided enhancement path as particles to obtain the initial particle swarm.

[0118] The optimal path in the initial particle swarm is determined based on the optimization objective, and the change in turning angle at each intermediate point on the optimal path is calculated. The calculation formula is as follows:

[0119]

[0120] Define a steering angle change threshold. If the steering angle change value of the intermediate point is greater than the steering angle change threshold, then mark the intermediate point and several adjacent intermediate points as the key optimization area.

[0121] By integrating all key optimization regions, the region to be optimized is obtained.

[0122] An adaptive particle swarm optimization (PSO) model is constructed, and iterative optimization is performed based on the adaptive PSO model and the region to be optimized until the algorithm converges. The position update formula for standard PSO is:

[0123]

[0124]

[0125] in, and The first The particle in the first The speed and position of the generation This is the optimal position in the particle's history. This is the best historical position for the group. It is a random number.

[0126] The adaptive particle swarm optimization model includes a dynamic parameter adjustment strategy, which adaptively adjusts the model parameters according to the iteration progress.

[0127] In this embodiment,

[0128] It should be noted that in the initial particle swarm, the main path and secondary paths provide practically feasible obstacle avoidance schemes; the fused path inherits the superior characteristics of multiple paths; and the guided enhancement path introduces global directional priors. By selecting key optimization regions, local optimization is achieved, effectively reducing the dimensionality of the optimization problem and decreasing computational complexity. Through the dynamic parameter adjustment strategy of the adaptive particle swarm optimization model, the algorithm possesses strong global exploration capabilities in the early stages. Larger Larger scale), later shifted to localized, refined development ( Smaller (Larger). At the same time, the speed limit is also dynamically adjusted with the iteration process: a larger speed is allowed in the early stage (2.5 units) to promote exploration, moderately reduced in the middle stage (1.5 units), and further tightened in the later stage (0.8 units) to ensure convergence accuracy.

[0129] To prevent over-optimization and getting trapped in local optima, an early stopping mechanism is introduced: if the improvement in the global optimal fitness is less than a threshold over several consecutive iterations. If the convergence is not achieved, the optimization process is terminated prematurely.

[0130] S5. Decode the globally optimal particle to obtain the final flight path.

[0131] It should be noted that the path is represented as a series of ordered coordinate points, which can be directly converted into a waypoint sequence that can be recognized by the UAV flight control system to guide the UAV's autonomous flight.

[0132] Example 2:

[0133] See Figure 2 As shown, this embodiment is a UAV path planning system based on a fusion algorithm, including:

[0134] The data acquisition and spatial modeling module is used to acquire the starting coordinates, ending coordinates, and obstacle information in the flight environment of the flight mission, and to construct a spatial model of the flight environment based on the starting coordinates, ending coordinates, and obstacle information in the flight environment of the flight mission.

[0135] The initial feasible path generation module is used to define path constraints and optimization objectives. Based on the flight environment spatial model, path constraints, and optimization objectives, it uses an improved fast expanding random tree algorithm to generate an initial feasible path set. The initial feasible path set includes a main path and several secondary paths.

[0136] The weighted fusion and guidance enhancement module is used to perform weighted fusion and guidance enhancement on the initial set of feasible paths to obtain the weighted fusion path and the guidance enhancement path.

[0137] The particle optimization module is used to construct an initial particle swarm based on an initial feasible path set, a weighted fusion path, and a guided enhancement path using the particle swarm optimization algorithm, and to iteratively optimize the initial particle swarm to obtain the globally optimal particle.

[0138] The final path generation module is used to decode the globally optimal particle to obtain the final flight path.

[0139] Example 3:

[0140] To verify the effectiveness of the proposed UAV path planning method based on fusion algorithm (P-NRRTC), comparative experiments were conducted with current advanced path planning algorithms in various complex static obstacle environments. The experimental environments covered two typical scenarios: multiple narrow passages and cluttered environments with multiple obstacles.

[0141] Comparison Algorithms: Three representative algorithms were selected as the comparison benchmarks: ; ;RRT-PSO.

[0142] Evaluation metrics: Fitness is calculated based on the optimization objective function as the core evaluation criterion. Path length, path tortuosity, algorithm execution time, and planning success rate are also statistically analyzed to comprehensively measure the path quality and planning efficiency of the algorithm.

[0143] Experimental setup: Each algorithm was run independently multiple times in each experimental environment, and the average value and standard deviation of each indicator were calculated to ensure the statistical reliability of the results.

[0144] (1) Multiple narrow channel model: such as Figure 3 As shown, the multi-narrow passage model environment is a 50×50 planar space, with the horizontal axis representing the X-coordinate of the two-dimensional plane and the vertical axis representing the Y-coordinate. Three vertical columnar obstacles are present, each divided into upper and lower sections with a 1-unit gap in between, starting at (3,25) and ending at (47,25). This is a typical narrow passage test scenario; the gap formed by the three columns constitutes the critical passage, suitable for testing the algorithm's ability to navigate in narrow spaces.

[0145] use Figure 3In the multi-narrow-channel model, the UAV must pass through three narrow gaps consecutively to complete the planning task. The channel width is much smaller than the environmental scale, and the passage area is height-restricted, making the path search process highly susceptible to random sampling failures. This is a typical test model for evaluating the search efficiency, stability, and path connectivity of sampling algorithms in narrow-channel environments. The experiments involved 100 independent loop trials of each algorithm, yielding the path results, as shown below. Figure 5-7 As shown, Figure 5-7 The red lines in the middle represent the final path. Figure 5 The yellow line is the guide line, and the green, blue, and purple lines are the three initial paths; Figure 6 The green lines represent the extended tree paths; Figure 7 The blue line represents the initial path, while the yellow, green, and purple lines represent the three initial paths, respectively. The final fitness values, along with metrics such as path length, smoothness, execution time, convergence speed, and success rate, are shown in Table 1 below.

[0146] Table 1. Results after 100 repetitions in the multi-narrow channel model.

[0147]

[0148] because The algorithm finds infeasible paths in extremely narrow environments, so they are not reflected in the path result graph.

[0149] As can be seen from the data in Table 1, With a success rate of only 53%, it indicates insufficient stability in narrow, continuous channels, making it prone to failure to complete planning due to sampling failures or premature convergence during optimization. Therefore, it was not included in subsequent performance comparisons. In contrast, both P-NRRTC and RRT-PSO maintained a 100% success rate, demonstrating stronger exploration capabilities in narrow environments.

[0150] Regarding path length, P-NRRTC achieves 47.02, roughly on the same order of magnitude as RRT-PSO, with minimal difference. This is because in multi-narrow-channel models, path length is primarily determined by channel geometry, limiting the optimization potential of each algorithm. In terms of execution time, P-NRRTC's average execution time is 1.48s, significantly better than RRT-PSO (approximately 8.40s). This result demonstrates that in multi-narrow-channel scenarios, P-NRRTC can complete path search and optimization in a shorter time. Regarding convergence speed, P-NRRTC's average number of iterations is 18.6, higher than RRT-PSO (1.0). RRT-PSO's convergence speed advantage stems from its rapid entry into the PSO optimization phase after obtaining the initial path, but this comes at the cost of higher computation time and energy consumption. In contrast, P-NRRTC achieves a more balanced performance between search depth and computational efficiency.

[0151] Considering all performance metrics, the fitness index shown in Table 1 indicates that P-NRRTC has an average value of 24.48, which is significantly better than RRT-PSO (26.06). This result demonstrates that under high-constraint environments such as multiple narrow channels, P-NRRTC can achieve better overall path planning results while maintaining a high success rate.

[0152] (2) Cluttered multi-obstacle model: such as Figure 4 As shown, a chaotic multi-obstacle model is used, with over 30 rectangular obstacles of varying sizes and shapes randomly distributed, including squares, rectangles, and narrow passages. The horizontal axis represents the X-coordinate of a two-dimensional plane, and the vertical axis represents the Y-coordinate. The starting point is (5,5), and the ending point is (95,95), simulating a complex real-world environment where the obstacles are randomly distributed, including narrow passages and open areas. This comprehensively tests the algorithm's adaptability in unstructured and complex environments.

[0153] use Figure 4 The model depicts a chaotic multi-obstacle environment, containing over 30 randomly distributed obstacles of varying shapes, simulating complex real-world environments with disordered obstacle distribution and narrow, irregular passages. This type of scenario poses significant challenges to the algorithm's global exploration capabilities, robustness to local optimization, and adaptability to unstructured environments. Experiments were conducted, with 100 independent loop trials performed on each algorithm to obtain the path results, as shown below. Figure 8-11 As shown, Figure 8-11 The red lines in the middle represent the final path. Figure 8 The yellow line is the guide line, and the green, blue, and purple lines are the three initial paths; Figure 9 The green lines represent the extended tree path, and the orange lines represent the ellipsoidal range. Figure 10The blue line represents the initial path, while the yellow, green, and purple lines represent three separate initial paths. Figure 11 The green lines represent the expanded tree paths. The final fitness values, along with metrics such as path length, smoothness, execution time, convergence speed, and success rate, are shown in Table 2 below.

[0154] Table 2. Results after 100 repetitions in the cluttered multi-obstacle model.

[0155]

[0156] In terms of average path length, P-NRRTC achieved a good result of 131.69±0.94, which is better than... (134.99±4.15) and (134.21±3.69), comparable to RRT-PSO (131.56±1.40). In such a complex environment, P-NRRTC still maintains a shorter path length, indicating that its multi-path fusion and guide point mechanism helps identify efficient passage areas in cluttered spaces. In terms of average execution time, P-NRRTC performs best, requiring only 6.6949±1.2916s, significantly lower than... (9.9285s) (19.1251s) and RRT-PSO (15.7895s). This verifies that P-NRRTC can maintain high computational efficiency in highly complex, unstructured environments. Regarding path smoothness, P-NRRTC has the best average smoothness of 2.19±0.35 among the four, significantly outperforming them. (4.05±1.13) and RRT-PSO (3.08±1.23), compared with (2.23±0.67) is comparable. This indicates that even in environments with randomly distributed obstacles and tortuous paths, P-NRRTC can still effectively improve the smoothness of the path and enhance the executability of the trajectory through the PSO optimization stage with local dynamic parameter tuning. In terms of convergence speed, the average number of iterations for P-NRRTC is 5.3, significantly lower than... (30.7) and (8.5) once again demonstrates its rapid convergence capability.

[0157] Considering all performance metrics, the average fitness of P-NRRTC is 68.29±0.59, which is the best among the four algorithms, further demonstrating its superior comprehensive planning performance in complex environments.

[0158] Example 4:

[0159] This embodiment provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0160] This computer device can be a server, and its internal structure diagram can be as follows: Figure 12 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores server data. The network interface communicates with external terminals via a network connection. When executed by the processor, the computer program implements a UAV path planning method based on a fusion algorithm.

[0161] Those skilled in the art will understand that Figure 12 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0162] Example 5:

[0163] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0164] If the functions implemented by the method are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art or the current technical solution, can be embodied in the form of a software product. This current computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0165] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0166] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0167] The technical features of this invention not described can be implemented by or using existing technology, and will not be repeated here. Of course, the above description is not a limitation of this invention, and this invention is not limited to the examples above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of this invention should also be within the protection scope of this invention.

Claims

1. A UAV path planning method based on a fusion algorithm, characterized in that, Includes the following steps: Obtain the starting coordinates, ending coordinates, and obstacle information in the flight environment of the flight mission, and construct a flight environment spatial model based on the starting coordinates, ending coordinates, and obstacle information in the flight environment. Define path constraints and optimization objectives. Based on the flight environment spatial model, path constraints, and optimization objectives, use an improved fast expanding random tree algorithm to generate an initial set of feasible paths. The initial set of feasible paths includes a main path and several secondary paths. The initial set of feasible paths is weighted and enhanced to obtain the weighted fused path and the enhanced path. The particle swarm optimization algorithm is adopted to construct an initial particle swarm based on the initial feasible path set, weighted fusion path and guided enhancement path, and to iteratively optimize the initial particle swarm to obtain the globally optimal particle. Decode the globally optimal particle to obtain the final flight path; The weighted fusion includes: The main path and several secondary paths are resampled to the same number of nodes using linear interpolation to obtain the aligned main path and several aligned secondary paths; The aligned main path and several aligned secondary paths are weighted and merged using random weighted interpolation to obtain a weighted merged path. The guidance enhancements include: A straight line is determined based on the starting point coordinates and the ending point coordinates as the sampling line; Samples are taken uniformly along the sampling line to obtain several sampling points; Based on the obstacle geometry model, collision detection is performed on each sampling point, and sampling points that do not collide are retained as guide points to obtain a set of guide points; Define a guide point insertion threshold. For each guide point, if the vertical distance from the guide point to the main path is less than the guide point insertion threshold, then insert the guide point into the main path to obtain the enhanced guide path.

2. The UAV path planning method based on fusion algorithm according to claim 1, characterized in that, The construction of the flight environment space model includes: The system acquires the starting coordinates, ending coordinates, and obstacle information in the flight environment for the flight mission; the obstacle information includes at least the location and size of the obstacles. The flight boundary is determined based on the starting point coordinates and the ending point coordinates. An initial obstacle geometric model is constructed based on obstacle information, and the initial obstacle geometric model is expanded to obtain the obstacle geometric model. A two-dimensional continuous space is constructed based on the geometric model of the flight boundary and obstacles to obtain the flight environment space model.

3. The UAV path planning method based on fusion algorithm according to claim 2, characterized in that, The path constraints include obstacle avoidance constraints and path tortuosity constraints. The optimization objective is to minimize the path length, path tortuosity, and algorithm execution time, as expressed by the following formula: in, For path length, For path tortuosity, For the algorithm execution time, These are the path length weight, path tortuosity weight, and algorithm execution time weight, respectively.

4. The UAV path planning method based on fusion algorithm according to claim 3, characterized in that, The generation of the initial feasible path set includes: Based on the flight environment spatial model, path constraints, and optimization objectives, an improved fast extended random tree algorithm is executed using a multi-threaded parallel computing method to generate several initial paths; Calculate the path length of several initial paths, and filter them according to the path length. The initial path with the shortest path length is selected as the main path, and the initial path with a path length greater than the shortest path length is selected as the secondary path. By integrating the main path and the secondary paths, an initial set of feasible paths is obtained.

5. The UAV path planning method based on fusion algorithm according to claim 4, characterized in that, The improved fast expanding random tree algorithm includes a target-oriented sampling strategy, a parent node selection strategy, and a search radius adjustment strategy; The target-oriented sampling strategy is to preset the target bias probability and use a random sampling method to sample target points based on the target bias probability. The parent node selection strategy includes evaluating candidate parent nodes within the search radius when expanding to a new node, and selecting the node with the minimum cumulative cost from the starting point to the new node as the parent node. The cumulative cost is calculated using the cumulative Euclidean distance, as expressed by the following formula: in, For the new node Centered on, search radius The set of all nodes within, The distance is Euclidean. The search radius adjustment strategy involves adjusting the search radius according to the size of the expanded tree, as expressed by the following formula: in, As an adjustment constant, The number of nodes in the tree. For spatial dimensions, As a multiple factor, To extend the step size.

6. The UAV path planning method based on a fusion algorithm according to claim 5, characterized in that, The iterative optimization of the initial particle swarm includes: The particle swarm optimization algorithm is used to encode the main path, slave path, weighted fusion path and guided enhancement path as particles to obtain the initial particle swarm. Determine the optimal path in the initial particle swarm based on the optimization objective, and calculate the change in turning angle at each intermediate point in the optimal path; Define a steering angle change threshold. If the steering angle change value of the intermediate point is greater than the steering angle change threshold, then mark the intermediate point and several adjacent intermediate points as the key optimization area. By integrating all key optimization regions, the region to be optimized is obtained. An adaptive particle swarm optimization model is constructed, and iterative optimization is performed based on the adaptive particle swarm optimization model and the region to be optimized until the algorithm converges. The adaptive particle swarm optimization model includes a dynamic parameter adjustment strategy, which adaptively adjusts the model parameters according to the iteration progress.

7. A UAV path planning system based on a fusion algorithm, characterized in that, The steps for performing the UAV path planning method based on the fusion algorithm as described in any one of claims 1 to 6 include: The data acquisition and spatial modeling module is used to acquire the starting coordinates, ending coordinates, and obstacle information in the flight environment of the flight mission, and to construct a spatial model of the flight environment based on the starting coordinates, ending coordinates, and obstacle information in the flight environment of the flight mission. The initial feasible path generation module is used to define path constraints and optimization objectives. Based on the flight environment spatial model, path constraints, and optimization objectives, it uses an improved fast expanding random tree algorithm to generate an initial feasible path set. The initial feasible path set includes a main path and several secondary paths. The weighted fusion and guidance enhancement module is used to perform weighted fusion and guidance enhancement on the initial set of feasible paths to obtain the weighted fusion path and the guidance enhancement path. The particle optimization module is used to construct an initial particle swarm based on an initial feasible path set, a weighted fusion path, and a guided enhancement path using the particle swarm optimization algorithm, and to iteratively optimize the initial particle swarm to obtain the globally optimal particle. The final path generation module is used to decode the globally optimal particle to obtain the final flight path.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the UAV path planning method based on the fusion algorithm as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the UAV path planning method based on the fusion algorithm as described in any one of claims 1 to 6.

Images