Unmanned aerial vehicle multi-target path planning method based on dynamic fitness adjustment

Abstract

The invention relates to the technical field of unmanned aerial vehicle intelligent control and path optimization, in particular to an unmanned aerial vehicle multi-target path planning method based on dynamic fitness adjustment, which comprises the following steps: topographic data acquisition and processing: acquiring data of a digital elevation model (DEM) and a digital orthoimage (DOM) through an unmanned aerial vehicle, generating a slope map and an obstacle density map, a dynamic fitness function and a dynamic weight adjustment mechanism are designed by combining factors such as terrain gradient, obstacle density and path length, and the weight of the fitness function is adjusted in real time according to different terrain characteristics so as to adapt to a variable environment. And the efficiency, the robustness and the adaptability of complex terrain path planning are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) intelligent control and path optimization, and in particular to a multi-objective path planning method for UAV based on dynamic fitness adjustment. Background Art

[0002] With the rapid development of drone technology, drones are increasingly used in fields such as agricultural spraying, resource inspection and disaster monitoring. However, drone path planning in complex terrain environments still faces many technical challenges, which are specifically reflected in the following aspects:

[0003] 1. Singleness of path optimization goal: Existing path planning methods mostly take the shortest path as the optimization goal, and rarely consider obstacle avoidance and energy consumption optimization in complex terrain. This optimization goal has certain applicability in flat terrain, but in terraced terrain with height differences, steep slopes and densely distributed obstacles, simply pursuing the shortest path may lead to increased task energy consumption and increased obstacle avoidance failure rate, reducing the overall task execution efficiency;

[0004] 2. Lack of dynamic adaptability: Traditional path planning algorithms (such as ant colony algorithms and genetic algorithms) usually use fixed fitness function weights, which makes it difficult to dynamically adjust the optimization target based on the local characteristics of complex terrain. For example, in terraced areas, steep slopes or areas with dense obstacles need to give priority to obstacle avoidance, while flat areas need to focus on optimizing path length. However, this static design makes it difficult for the algorithm to meet the demand for path flexibility in complex scenarios;

[0005] 3. Low efficiency of regional switching: Complex terrain often contains multiple partitions with significant height differences, such as different levels of terraces. When the drone switches paths between regions, if the energy consumption caused by path smoothness and height changes is not fully considered, it will lead to longer mission time and increased energy consumption. In addition, the lack of coordination between local partition optimization and global path planning is also a bottleneck in existing technologies.

[0006] 4. Insufficient use of complex terrain features: Although modern drones are equipped with high-precision sensors and cameras that can obtain data such as digital elevation models (DEMs) and digital orthophotos (DOMs), the existing technologies still make relatively rough use of these terrain features. For example, the role of slope maps and obstacle density maps in path planning is often limited to assisting path evaluation and is not deeply integrated into optimization algorithms, thus limiting planning efficiency.

[0007] 5. Limitations of hybrid optimization algorithms: Although existing hybrid optimization algorithms attempt to combine genetic algorithms and ant colony algorithms, they usually fail to effectively use terrain data for weight adjustment, resulting in limited global and local optimization capabilities for path planning. For complex terrain, these algorithms fail to dynamically adjust the weights of global search and local optimization, and cannot ensure that the path is optimal in terms of energy consumption and obstacle avoidance.

[0008] Therefore, there is an urgent need for a path planning method that can dynamically adapt to complex terrain characteristics and achieve multi-objective balanced optimization among path length, obstacle avoidance capability, and energy consumption, so as to improve the mission execution effect of UAVs in complex environments. Summary of the invention

[0009] The present invention provides a multi-target path planning method for an unmanned aerial vehicle based on dynamic fitness adjustment, which can effectively solve the problems raised in the above background technology.

[0010] To achieve the above object, the present invention provides the following technical solution: a multi-target path planning method for an unmanned aerial vehicle based on dynamic fitness adjustment, comprising the following steps:

[0011] Terrain data processing: Use drones to obtain digital elevation model (DEM) and digital orthophoto (DOM) data, generate slope maps and obstacle density maps, and divide the terrain into different regions, marking the fitness priorities of different regions;

[0012] Fitness function construction: Combine terrain slope, obstacle density and path length to construct a multi-objective fitness function. The fitness function weight is dynamically adjusted according to the terrain characteristics:

[0013] In flat areas, increase the weight of the path length factor;

[0014] In steep slope areas, increase the weight of obstacle avoidance factors;

[0015] In areas with dense obstacles, reduce the weight of the path length factor;

[0016] The improved hybrid optimization algorithm path planning includes the following processes:

[0017] Initial path generation: Genetic algorithm is used to quickly generate the initial path;

[0018] Local path optimization: Combine the improved ant colony algorithm to optimize the path selection details;

[0019] Optimized path output: Generates path node sequence and optimization indicators, including path length, obstacle avoidance success rate and energy consumption performance.

[0020] According to the above technical solution, the fitness function is constructed by combining multi-dimensional factors such as terrain slope, obstacle density and path length to construct a multi-objective fitness function, the formula is:

[0021] f total =ω1·f slop e+ω2·f obstacle +ω3·f length

[0022] in:

[0023] f slope : slope fitness factor;

[0024] f obstacle : obstacle density fitness factor;

[0025] f length : Path length fitness factor;

[0026] ω1, ω2, ω3: dynamically adjusted weights.

[0027] According to the above technical solution, the dynamic path selection probability is determined by the pheromone concentration and the fitness value. The specific calculation formula is as follows:

[0028]

[0029] in:

[0030] τ ij : pheromone concentration on the path;

[0031] η ij : The fitness value of the path;

[0032] α, β: importance parameters for adjusting pheromone and fitness values.

[0033] According to the above technical solution, the initial path generation of the genetic algorithm includes: encoding the terrain nodes to generate chromosomes;

[0034] Combine the normalization of total path length and obstacle avoidance rate to construct a multi-objective fitness function;

[0035] Combined with multi-objective fitness values, the weights are dynamically adjusted to enhance the global search capability of the genetic algorithm;

[0036] The initial path sequence is generated through crossover, mutation, and selection operations.

[0037] According to the above technical solution, the hybrid optimization algorithm path planning includes: generating an initial path using a genetic algorithm to provide a global search capability;

[0038] The initial path is locally optimized by using the improved ant colony algorithm to avoid falling into the local optimum.

[0039] The global search advantage of genetic algorithm and the local optimization ability of ant colony algorithm are combined to generate the final optimization path.

[0040] According to the above technical solution, the pheromone initialization includes the following steps: in the initial stage of path planning, the pheromone concentration of the path node is dynamically allocated according to the fitness graph, and the path with a high fitness value is preferentially allocated with a higher concentration to enhance the inspiration of the algorithm.

[0041] According to the above technical solution, the dynamic update rules of the pheromone include: updating the path fitness optimization result;

[0042] Combine the path fitness optimization results and volatilization mechanism to dynamically adjust the pheromone concentration of the path nodes; Strengthening and attenuation mechanism: prioritize strengthening the pheromone concentration of high-fitness paths to guide ants to prioritize high-quality paths;

[0043] At the same time, the pheromone concentration of low fitness paths is attenuated to reduce the interference of bad paths.

[0044] According to the above technical solution, the partition optimization strategy of the path planning includes: optimizing the path in the local area to avoid obstacles and steep slopes; when switching paths between areas, giving priority to paths with shorter path lengths and smaller height differences;

[0045] Reducing energy consumption in path switching by adjusting the strategy across regions.

[0046] According to the above technical solution, the weight of each factor in the fitness function can be adjusted according to actual needs to meet the planning needs of different complex terrain scenarios.

[0047] According to the above technical scheme, the method is applicable to the drone spraying, inspection and data collection tasks in terraced agriculture, and can also be extended to mountain inspection and disaster monitoring scenarios;

[0048] The dynamic adjustment mechanism dynamically optimizes the fitness weight according to the terrain characteristics of different areas, combined with offline analysis and preset rules, thereby improving the adaptability and flexibility of path planning.

[0049] Compared with the prior art, the invention has the following beneficial effects: the invention has a scientific and reasonable structure, is safe and convenient to use,

[0050] Strong adaptability: In view of the characteristics of complex terrain, the present invention constructs a multi-objective fitness function and dynamically adjusts the adaptive weights to enable the UAV to respond to the changing needs of different terrains in real time, significantly improving the flexibility of path planning.

[0051] Efficient hybrid optimization algorithm: The present invention combines the genetic algorithm with the improved ant colony algorithm, which not only utilizes the advantages of the genetic algorithm in global path search, but also combines the efficiency of the ant colony algorithm in local path optimization, and finally realizes high-quality path planning.

[0052] Wide range of application scenarios: The present invention is suitable for multiple complex terrain scenarios such as terraced agricultural spraying, mountain inspection, disaster monitoring, etc., and has good versatility and practicality. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0054] In the attached picture:

[0055] Figure 1 This is the overall flow chart of the path planning of the present invention;

[0056] Figure 2 A flow chart for terrain data processing and fitness function generation of the present invention;

[0057] Figure 3 This is a flow chart of the algorithm path planning of the present invention. DETAILED DESCRIPTION

[0058] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0059] Embodiment 1:

[0060] like Figure 1 As shown, the present invention provides a technical solution, a multi-target path planning method for an unmanned aerial vehicle based on dynamic fitness adjustment, comprising the following main steps:

[0061] Terrain data acquisition and processing: The laser radar and high-definition camera equipment carried by the drone are used to obtain the digital elevation model DEM and digital orthophoto DOM data of the target area, generate slope maps and obstacle density maps, and divide the terrain into regional zones.

[0062] Fitness function construction: Taking into account the path length, obstacle avoidance success rate and energy consumption factors, a multi-objective fitness function is constructed. Each factor has a dynamically adjustable weight in the fitness function, which is adjusted in real time according to the terrain characteristics.

[0063] Dynamic weight adjustment mechanism: In flat areas, the weight of the path length factor is increased, and the path with the shortest path length is given priority; in steep slope areas, the weight of the obstacle avoidance factor is increased to ensure the safety of the drone path; in areas with dense obstacles, the weight of the path length factor is reduced to ensure safety. The weight adjustment process combines offline analysis and preset rules to dynamically optimize the fitness weight.

[0064] Improved hybrid optimization algorithm path planning: First, the initial path is generated by genetic algorithm, and then the path is locally refined and optimized in combination with the improved ant colony algorithm to ensure that the path planning has both global search capabilities and local optimization effects.

[0065] Optimized path output: Finally, the optimized path node sequence is generated, and optimization indicators such as path length, obstacle avoidance success rate and energy consumption are calculated. The results are output to the drone control system to achieve precise flight.

[0066] Embodiment 2:

[0067] Terrain data acquisition and processing

[0068] like Figure 2 As shown in Figure 1, the process of acquiring and processing terrain data includes the following detailed steps:

[0069] 3D terrain data collection: Use drones equipped with high-precision cameras to take aerial photos of the target area and collect multi-view image data to ensure coverage of the entire planned area. Strictly control flight altitude and flight overlap rate to ensure that the lateral overlap rate is not less than 70% and the longitudinal overlap rate is not less than 60% to ensure the comprehensiveness and accuracy of terrain data.

[0070] 3D modeling and DEM generation: Use professional software such as Pix4D or Metashape to perform 3D reconstruction on the collected image data, generate a point cloud model of the terrain, extract terrain features from the point cloud, and generate a digital elevation model (DEM) to describe the elevation of each terrain location.

[0071] DOM generation: Orthorectify the image data, eliminate angular distortion, and generate distortion-free digital orthophoto DOM for subsequent object recognition and obstacle analysis.

[0072] Generation of slope map and obstacle density map: The slope value of each pixel is calculated through DEM to divide the terrain into flat land, gentle slope and steep slope areas; the density of obstacles (such as vegetation and buildings) per unit area is analyzed based on DOM data to generate an obstacle density map.

[0073] Embodiment 3:

[0074] Dynamic adjustment mechanism of fitness function

[0075] like Figure 2 As shown in Figure 2, the dynamic adjustment mechanism of the fitness function is implemented through the following steps:

[0076] Fitness function construction: The fitness function of the present invention combines the slope factor, obstacle density factor and path length factor to comprehensively optimize multiple goals of path planning. The specific fitness function calculation formula is as follows:

[0077] f total =ω1·f slope +ω2·f obstacle +ω3·f length

[0078] in:

[0079] f slope : Slope adaptability factor, which indicates the influence of terrain slope on path planning. The slope factor will be dynamically adjusted according to different terrains (such as flat land, gentle slope, steep slope). The slope factor weight in the steep slope area will increase, while the slope factor weight in the flat area will be smaller.

[0080] f obstacle : Obstacle density fitness factor, which indicates the priority of avoiding obstacles. In areas with dense obstacles, the weight of the obstacle factor is higher, while in areas without obstacles, the weight of the obstacle factor is lower.

[0081] f length : Path length fitness factor, indicating the optimization requirement of the total path length.

[0082] In each iteration of path planning, the fitness weights are dynamically adjusted according to the terrain characteristics:

[0083] Flat land: The path length factor has a higher weight (ω3>ω2,ω1);

[0084] Steep slope: The obstacle avoidance factor has a higher weight (ω2>ω3,ω1);

[0085] Obstacle-dense areas: The weight of the obstacle factor is significantly increased, and the weight of the path length factor is reduced.

[0086] Dynamic adjustment formula: The dynamic adjustment weight formula is as follows:

[0087]

[0088] Among them, Vω i is the weight change, which is determined by the path planning status and terrain characteristics.

[0089] The following specific implementation method provides a detailed description of the key technologies of the present invention, and gradually explains the specific implementation method of the technical solution in combination with actual application scenarios to ensure the technical feasibility and effect of the implementation.

[0090] Embodiment 4:

[0091] Hybrid algorithm path planning

[0092] like Figure 3 As shown, the hybrid algorithm path planning process includes the following steps:

[0093] Initial path generation: Genetic algorithms are used to encode terrain nodes, generate initial chromosomes, and generate initial path sequences through crossover, mutation, and selection operations.

[0094] Local path optimization: After the initial path is generated, the path is locally optimized in combination with the improved ant colony algorithm, and the details of the path are gradually improved by using the dynamic allocation and updating of pheromones.

[0095] Improved ant colony algorithm path planning:

[0096] Pheromone initialization: The pheromone concentration of the path nodes is initialized according to the fitness graph. High fitness areas are given priority to higher initial pheromone concentrations to guide the drone to prioritize paths in these areas. Through this initialization mechanism, the overall inspiration of path planning is improved. For different terrain characteristics, the specific distribution of initial pheromones is as follows: The initial pheromone concentration in high fitness areas is twice that of ordinary areas.

[0097] Path selection rules: During the path selection process, ants select paths based on pheromone concentration and fitness value. The priority of path selection is dynamically adjusted according to the terrain characteristics. For example, in steep slopes and areas with dense obstacles, the weight of the obstacle avoidance factor will increase when selecting a path, reducing the probability of selecting these high-risk areas. In flat areas, the path length factor will be given priority to increase the efficiency of the drone flight. The specific calculation is as follows:

[0098]

[0099] Where: τ ij is the pheromone concentration on the path, η ij is the fitness value of the path, α and β are the importance parameters for adjusting pheromones and fitness values. According to different terrain characteristics, the weight of the obstacle avoidance factor is dynamically increased in steep slopes and obstacle-dense areas.

[0100] Pheromone update rules: During the pheromone update process, the details of the reinforcement and attenuation mechanisms are as follows:

[0101] Reinforcement mechanism: Prioritize strengthening the pheromone concentration of paths with higher fitness to guide more ants to choose high-quality paths.

[0102] Dynamic evaporation rate: The evaporation rate of pheromones is no longer fixed, but is adjusted according to terrain and path fitness feedback. In the early stage of iteration, a high evaporation rate is maintained to increase the diversity of path exploration; in the later stage, the evaporation rate is reduced to focus on known high-quality paths.

[0103] Compared with the traditional ant colony algorithm:

[0104] Difference in pheromone update mechanism: The traditional ant colony algorithm does not update the pheromone concentration in the height difference area sufficiently, which makes the path planning easy to fall into the local optimum. The present invention significantly improves this by combining the dynamic volatilization rate and regional characteristics.

[0105] Difference in path selection: The traditional ant colony algorithm mainly selects paths based on fixed fitness weights, while the improved algorithm combines real-time terrain characteristics and dynamically adjusts weights, making path selection under different terrains more flexible and reasonable.

[0106] Hybrid optimization process: Dynamically adjust the weight parameters of the genetic algorithm and the ant colony algorithm to adapt to different terrain characteristics, ensuring that the path has both global search capabilities and local fine optimization.

[0107] Example 5: Local and global path planning

[0108] Local path planning: Figure 3 As shown in the figure, within the terraced area, a path that avoids obstacles and has low energy consumption is generated through the overlay analysis of the slope map and the obstacle density map. The specific path planning includes extracting the elevation data of the local area, calculating the path selection probability in combination with the fitness function, and gradually optimizing the pheromone distribution of the path nodes.

[0109] Global path planning: When the UAV switches across regions, it first analyzes the height difference and obstacle distribution between the two regions, and selects the optimal switching path in combination with the dynamic fitness adjustment mechanism, giving priority to switching nodes with smaller height differences and higher obstacle avoidance rates to ensure path smoothness and lower energy consumption.

[0110] Path planning result output: The optimal path finally generated includes a combination of local and global paths. Through this strategy, energy consumption can be effectively reduced and smooth cross-regional path switching can be achieved.

[0111] Finally, it should be noted that the above description is only a preferred example of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A multi-target path planning method for unmanned aerial vehicles based on dynamic fitness adjustment, characterized in that: The following steps are involved: Terrain data processing: Use drones to obtain digital elevation model (DEM) and digital orthophoto (DOM) data, generate slope maps and obstacle density maps, and divide the terrain into different regions, marking the fitness priorities of different regions; Fitness function construction: Combine terrain slope, obstacle density and path length to construct a multi-objective fitness function. The fitness function weight is dynamically adjusted according to the terrain characteristics: in flat areas, the weight of the path length factor is increased; In steep slope areas, increase the weight of obstacle avoidance factors; In areas with dense obstacles, reduce the weight of the path length factor; The improved hybrid optimization algorithm path planning includes the following processes: Initial path generation: Genetic algorithm is used to quickly generate the initial path; Local path optimization: Combine the improved ant colony algorithm to optimize the path selection details; Optimized path output: Generates path node sequence and optimization indicators, including path length, obstacle avoidance success rate and energy consumption performance.

2. The multi-objective path planning method for unmanned aerial vehicle based on dynamic fitness adjustment according to claim 1 is characterized in that: The fitness function is constructed by combining the multi-dimensional factors of terrain slope, obstacle density and path length to construct a multi-objective fitness function, the formula is: f total =ω1·f slope +ω2·f obstacle +ω3·f length in: f slope : slope fitness factor; f obstacle : obstacle density fitness factor; f length : Path length fitness factor; ω1, ω2, ω3: dynamically adjusted weights.

3. The multi-objective path planning method for unmanned aerial vehicle based on dynamic fitness adjustment according to claim 1 is characterized in that: The dynamic path selection probability is determined by the pheromone concentration and fitness value. The specific calculation formula is as follows: in: τ ij : pheromone concentration on the path; η ij : The fitness value of the path; α, β: importance parameters for adjusting pheromone and fitness values.

4. The method for multi-objective path planning of an unmanned aerial vehicle based on dynamic fitness adjustment according to claim 3 is characterized in that: The initial path generation of the genetic algorithm includes encoding the terrain nodes to generate chromosomes; Combine the normalization of total path length and obstacle avoidance rate to construct a multi-objective fitness function Combined with multi-objective fitness values, the weights are dynamically adjusted to enhance the global search capability of the genetic algorithm; The initial path sequence is generated through crossover, mutation, and selection operations.

5. The method for multi-objective path planning of an unmanned aerial vehicle based on dynamic fitness adjustment according to claim 4 is characterized in that: The hybrid optimization algorithm path planning includes: generating an initial path using a genetic algorithm to provide a global search capability; The initial path is locally optimized by using the improved ant colony algorithm to avoid falling into the local optimum. The global search advantage of genetic algorithm and the local optimization ability of ant colony algorithm are combined to generate the final optimization path.

6. The method for multi-objective path planning of an unmanned aerial vehicle based on dynamic fitness adjustment according to claim 5, characterized in that: The pheromone initialization includes the following steps: in the initial stage of path planning, the pheromone concentration of the path node is dynamically allocated according to the fitness graph, and the path with a high fitness value is preferentially allocated with a higher concentration.

7. The method for multi-objective path planning of an unmanned aerial vehicle based on dynamic fitness adjustment according to claim 5, characterized in that: The dynamic update rules of the pheromone include: updating the path fitness optimization results; Combine the path fitness optimization results and volatilization mechanism to dynamically adjust the pheromone concentration of the path nodes; Strengthening and attenuation mechanism: prioritize strengthening the pheromone concentration of high-fitness paths to guide ants to prioritize high-quality paths; At the same time, the pheromone concentration of low fitness paths is attenuated to reduce the interference of bad paths.

8. The method for multi-objective path planning of an unmanned aerial vehicle based on dynamic fitness adjustment according to claim 1, characterized in that: The partition optimization strategy of the path planning includes: optimizing the path in a local area to avoid obstacles and steep slopes; when switching paths between areas, giving priority to paths with shorter path lengths and smaller height differences; Reducing energy consumption in path switching by adjusting the strategy across regions.

9. The method for multi-objective path planning of an unmanned aerial vehicle based on dynamic fitness adjustment according to claim 1, characterized in that: The weights of each factor in the fitness function can be adjusted according to actual needs.

10. The method for multi-objective path planning of an unmanned aerial vehicle based on dynamic fitness adjustment according to claim 1, characterized in that: The method is suitable for UAV spraying, inspection and data collection tasks in terraced agriculture, and can also be extended to mountain inspection and disaster monitoring scenarios; The dynamic adjustment mechanism dynamically optimizes the fitness weight based on the terrain characteristics of different areas, combined with offline analysis and preset rules.

Images