CN-121384039-B - Unmanned aerial vehicle path planning method and system suitable for three-dimensional terrain

CN121384039BCN 121384039 BCN121384039 BCN 121384039BCN-121384039-B

Abstract

The invention belongs to the technical field of unmanned plane path planning. The unmanned aerial vehicle path planning method and system suitable for three-dimensional terrain are provided, aiming at the problems of two-dimensional limitation, low search efficiency, easy-to-trap local optimum and the like existing in the process of unmanned aerial vehicle path planning by a traditional ant colony algorithm, Q-Learning is integrated into the ant colony algorithm based on DEM modeling, heuristic functions and pheromone strategies are optimized, the optimal path is integrated and smooth processing is carried out, the efficiency and the safety of path planning under complex terrain are improved, and the unmanned aerial vehicle path planning method and system are suitable for mountain area inspection, agricultural plant protection and other scenes.

Inventors

ZHOU JUN
Chai Kunqing
LI SHOUYE
MA YONGXIN
LI MINGMING
YAO YUBO

Assignees

山东大学

Dates

Publication Date: 20260512
Application Date: 20251223

Claims (10)

1. The unmanned aerial vehicle path planning method suitable for the three-dimensional terrain is characterized by comprising the following steps of: Modeling the topography of the planning space based on the digital elevation model to obtain a planning space model, constructing an objective function with minimum path length and threat, and obtaining an initial Q value, a pheromone weight, a heuristic information scaling factor, a Q value scaling factor, an initial pheromone volatilization value, a learning rate and a discount factor; based on a Q-Learning algorithm, constructing a heuristic function model, and integrating an initial Q value, a pheromone weight, a heuristic information scaling factor and a Q value scaling factor to obtain an improved state transition formula; Based on the objective function, determining ants with path lengths smaller than the current iteration average path length as elite ants, updating the pheromone concentration of the paths according to an pheromone increment equation, and simultaneously combining the initial pheromone volatilization value and the current iteration times, and dynamically adjusting the pheromone volatilization value according to a self-adaptive pheromone fluctuation factor equation; According to the improved state transition formula and the pheromone concentration, ants select the next node from the allowed node set meeting the collision and turning angle constraint, and record the path length, threat coefficient, accumulated benefits and node access times; Calculating the maximum cumulative profit frequency based on the cumulative profit and the node access times, and updating the Q value among the nodes according to a Q value updating formula by combining the learning rate and the discount factor; After each generation of ants completes the path planning task, selecting three ants with optimal path length, threat coefficient and accumulated benefit combination according to the objective function and the updated Q value among nodes, and fusing local paths of the three ants to obtain a fused path; And adopting a third-order Bezier curve to carry out smoothing treatment on the fusion path, and generating an optimal three-dimensional path meeting the constraints of collision and turning angles in the planning space model.
2. The unmanned aerial vehicle path planning method of claim 1, wherein the unmanned aerial vehicle path planning method comprises the steps of, The heuristic function model is: ; Wherein, the For the current node To the next node Is used for the distance between euclidean distance(s), Is the end point To the next node Is used for the distance between euclidean distance(s), To search for the expected degree on the path node.
3. The unmanned aerial vehicle path planning method for three-dimensional terrain according to claim 2, wherein, The improved state transition formula is as follows: ; Wherein, the Is an ant From the current node To the next node Is used for the probability transfer equation of (2), Is the weight of the pheromone, Is a heuristic weight, which is a weight, Representing ants Can search for nodes at the current time Where it reaches the next set of nodes At the position of the first part, Is the pheromone concentration value at the search path node, Is the degree of expectation on the nodes of the search path, Is a value scaling factor; Is the current node To the next node Is the accumulation of (2) A value; Representing the moment.
4. The unmanned aerial vehicle path planning method of claim 1, wherein the unmanned aerial vehicle path planning method comprises the steps of, The pheromone delta equation is: ; Wherein, the Is an ant In the first place The pheromone that is left over is iterated, Is the constant of the pheromone, Is an ant The length of the path to be searched for, Is the average path length at this time of iteration, Is the shortest path length at this time of iteration, Is an adaptive weight value that is used to determine, Representing the total number of ants.
5. The unmanned aerial vehicle path planning method of claim 4, wherein the unmanned aerial vehicle path planning method comprises the steps of, Adaptive weight value Comprising: 。
6. the unmanned aerial vehicle path planning method of claim 1, wherein the unmanned aerial vehicle path planning method comprises the steps of, The adaptive pheromone wave factor equation is: ; Wherein, the Representing the normal distribution factor to be introduced, Representing the initial pheromone volatilization value, Representing the number of current iterations, Is the threshold value of the number of iterations, Represents the normal distribution scale parameter of the distribution, Representing normal distribution position parameters.
7. The unmanned aerial vehicle path planning method of claim 1, wherein the unmanned aerial vehicle path planning method comprises the steps of, Three ants with optimal combinations of path length, threat coefficients and cumulative benefit are selected using trade-off factors: ; Wherein, the And Is a weight parameter that is used to determine the weight of the object, Is the cumulative benefit of ants traveling on a path, Is a threatening item and is presented with a code, Is a distance item that is to be used, Representing a terrain threat assessment function, And a fitness value representing the flight distance.
8. Unmanned aerial vehicle path planning system suitable for three-dimensional topography, characterized by comprising: The modeling initialization unit is configured to model the topography of the planning space based on the digital elevation model according to the acquired topography data to obtain a planning space model, construct an objective function with minimum path length and threat, and acquire an initial Q value, a pheromone weight, a heuristic information scaling factor, a Q value scaling factor, an initial pheromone volatilization value, a learning rate and a discount factor; the transfer formula improving unit is configured to construct a heuristic function model based on a Q-Learning algorithm, integrate an initial Q value, a pheromone weight, a heuristic information scaling factor and a Q value scaling factor and obtain an improved state transfer formula; The pheromone updating unit is configured to judge ants with path lengths smaller than the current iteration average path length as elite ants based on the objective function, update the pheromone concentration of the paths according to an pheromone increment equation, and dynamically adjust the pheromone volatilization value according to a self-adaptive pheromone fluctuation factor equation by combining the initial pheromone volatilization value and the current iteration times; A path search recording unit configured to record path length, threat coefficient, accumulated benefits and node access times according to the improved state transition formula and the pheromone concentration, and the ants select the next node from the allowed node set meeting the collision and turning angle constraint; A Q value updating unit configured to calculate a maximum cumulative benefit frequency based on the cumulative benefit and the number of node accesses, and update the Q value between nodes according to a Q value updating formula in combination with the learning rate and the discount factor; The path fusion unit is configured to screen three ants with minimum path length, minimum threat coefficient and optimal accumulated benefit combination according to the objective function and the updated Q value between nodes, extract local paths, and integrate the local paths through path distance and benefit balance formulas by utilizing balance factors; and the path smoothing unit is configured to carry out smoothing processing on the fusion path by adopting a third-order Bezier curve to generate an optimal three-dimensional path meeting the collision and turning angle constraint in the planning space model.
9. A computer device comprises a processor and a computer-readable storage medium; a processor adapted to execute a computer program; A computer readable storage medium having stored therein a computer program which, when executed by the processor, implements the unmanned aerial vehicle path planning method of any one of claims 1 to 7 adapted to three-dimensional terrain.
10. A computer readable storage medium, characterized in that it stores a computer program adapted to be loaded by a processor and to perform the unmanned aerial vehicle path planning method according to any of claims 1 to 7, which is applicable to three-dimensional terrain.

Description

Unmanned aerial vehicle path planning method and system suitable for three-dimensional terrain Technical Field The invention relates to the technical field of unmanned aerial vehicle path planning, in particular to an unmanned aerial vehicle path planning method and system suitable for three-dimensional terrain. Background The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art. With the rapid development of unmanned aerial vehicle technology, the unmanned aerial vehicle is increasingly widely applied to the fields of mountain area inspection, agricultural plant protection, electric power line inspection and the like. In these applications, achieving efficient, safe three-dimensional full-coverage path planning is a key technical challenge. As the drone performs the overlay task in three-dimensional space, accurate position data must be constantly retrieved from the original terrain model. Flying at constant altitude limits their effectiveness in covering complex terrain (e.g., mountainous and valley areas) because safety restrictions prevent them from coming closer to the earth's surface. The unmanned aerial vehicle is optimized according to climbing capacity, sensor capacity and terrain features of the area, and the three-dimensional flight path of the unmanned aerial vehicle is adjusted accordingly. The ant colony algorithm (ACO, ant Colony Optimization) mimics the foraging behavior of ants in nature, seeking an optimal path through the positive feedback mechanism of pheromones. It also has some limitations in solving the path planning problem. When the ACO algorithm executes a path planning task, the key point is that each path node is selected according to the concentration of pheromone, after each generation of ants complete path searching, the pheromone is released to all paths through which the ants pass, even if only part of the paths are optimal, the path planning time is increased, the convergence speed to the optimal paths is reduced, each generation of the ants is updated, the feedback effect is not obvious enough, in addition, the ants can be blocked on some obstacle nodes, so that the optimization capability is reduced, and when the global optimal path appears later, the concentration of the pheromone on the current optimal path is too high, so that the population is ignored, the local optimal is trapped, the searching efficiency is reduced, and the premature convergence is caused, so that the conventional ant algorithm cannot effectively adapt to the path planning requirement of the three-dimensional terrain unmanned plane. Disclosure of Invention In order to solve the defects of the prior art, the invention provides an unmanned aerial vehicle path planning method and system suitable for three-dimensional terrain, which combines terrain data, models the terrain in a planning space based on a digital elevation model, expresses the three-dimensional unmanned aerial vehicle path planning problem as a constraint optimization problem with path length and threat as objective functions and with collision and turning angles as constraints, combines reinforcement learning and ant colony algorithm in a three-dimensional environment, and realizes unmanned aerial vehicle path planning with higher precision. In order to achieve the above purpose, the present invention adopts the following technical scheme: In a first aspect, the present invention provides a method for unmanned aerial vehicle path planning suitable for three-dimensional terrain. An unmanned aerial vehicle path planning method suitable for three-dimensional terrain comprises the following steps: Modeling the topography of the planning space based on the digital elevation model to obtain a planning space model, constructing an objective function with minimum path length and threat, and obtaining an initial Q value, a pheromone weight, a heuristic information scaling factor, a Q value scaling factor, an initial pheromone volatilization value, a learning rate and a discount factor; based on a Q-Learning algorithm, constructing a heuristic function model, and integrating an initial Q value, a pheromone weight, a heuristic information scaling factor and a Q value scaling factor to obtain an improved state transition formula; Based on an objective function, determining ants with path lengths smaller than the current iteration average path length as elite ants, updating the pheromone concentration of the paths according to an pheromone increment equation, and simultaneously combining an initial pheromone volatilization value and the current iteration times, and dynamically adjusting the pheromone volatilization value according to a self-adaptive pheromone fluctuation factor equation; According to the improved state transition formula and the pheromone concentration, ants select the next node from the allowed node set meeting the collision and turning angle