Unmanned aerial vehicle three-dimensional path planning method based on improved artificial travel mouse optimization algorithm

Abstract

The invention discloses an unmanned aerial vehicle three-dimensional path planning method based on an improved artificial travel mouse optimization algorithm, belongs to the technical field of unmanned aerial vehicle path planning methods, and aims at enabling an unmanned aerial vehicle to autonomously navigate, avoid obstacles, find an optimal path and improve task execution efficiency and safety in a complex environment. A constraint condition and a target function of unmanned aerial vehicle three-dimensional path planning are mapped into a search space and a fitness function of an optimization algorithm, an optimal flight path is searched by adopting an artificial travel mouse optimization algorithm, and optimization calculation is performed on the path in combination with unmanned aerial vehicle performance constraints, topographic features and meteorological conditions; aiming at a three-dimensional path planning problem of an unmanned aerial vehicle under a complex condition, an artificial travel mouse algorithm is adaptively improved, and algorithm parameters and structures are respectively adjusted according to mathematical characteristics of a problem model in the aspects of initialization, exploration and development; the flight path is adjusted based on environment change and self state information sensed by a sensor in real time, and the safety and task requirements are met.

Description

Technical Field

[0001] This invention relates to the field of UAV path planning methods, specifically to a UAV three-dimensional path planning method based on an improved artificial lemming optimization algorithm. Background Technology

[0002] In modern society, unmanned aerial vehicle (UAV) technology has been widely used in many fields such as agriculture, logistics, surveying and mapping and disaster relief. The autonomous navigation capability of UAVs has become the key to their widespread application, and path planning of UAVs is the core of achieving autonomous navigation and efficient operation.

[0003] Unmanned aerial vehicle (UAV) path planning refers to providing UAVs with an efficient, safe, and intelligent flight path, enabling them to perform various tasks in complex environments. This improves UAVs' navigation capabilities in complex environments, reduces flight distance and energy consumption, and enhances flight efficiency and economy. This process needs to consider multiple factors, including environmental obstacles, flight distance, flight time, energy consumption, and specific mission requirements.

[0004] However, existing path planning algorithms often fail to achieve ideal results when considering various complex factors. Furthermore, in real-world environments, dynamic events such as moving targets and weather changes can make it difficult for drones to perform tasks as planned. Therefore, existing path planning methods struggle to achieve satisfactory results in practical applications. Summary of the Invention

[0005] This invention proposes a UAV 3D path planning method based on an improved artificial lemming optimization algorithm. The UAV 3D path planning method based on an improved artificial lemming optimization algorithm described in this invention is as follows: A system model is constructed, environmental data is acquired using airborne sensors, and the environmental data is preprocessed to form a three-dimensional flight environment model of the UAV in combination with mission requirements. A problem-solving framework is established, mapping the constraints and objective function of UAV 3D path planning to the search space and fitness function of the optimization algorithm. The artificial lemming optimization algorithm is used to search for the optimal flight path, and the path is optimized by combining UAV performance constraints, terrain features and meteorological conditions. Algorithm adaptability optimization was carried out. For the three-dimensional path planning problem of UAVs under complex conditions, the artificial lemming algorithm was adaptively improved. The algorithm parameters and structure were adjusted according to the mathematical characteristics of the problem model in the initialization, exploration and development stages. Real-time dynamic path replanning adjusts the flight path based on real-time environmental changes and the drone's own status information perceived by sensors during flight, in order to meet safety and mission requirements.

[0006] Furthermore, in one embodiment of the present invention, the three-dimensional flight environment model specifically refers to: Multi-source environmental data is collected using sensors, and combined with the UAV's current location and mission requirements, a 3D flight environment model of the UAV is constructed. The objective function is expressed as: ; in, As weight, For path length cost, For the cost of a high standard deviation, Cost of planning path smoothness For the total cost; The model of the ground and obstacles is represented as follows: ; in, and These are the horizontal and vertical coordinates in three-dimensional space, used to locate the drone's position on the horizontal plane. The vertical height of the terrain and obstacles is used to mathematically simulate complex 3D terrain and obstacle distributions to detect whether the path intersects with the ground or obstacles.

[0007] Furthermore, in one embodiment of the present invention, the artificial lemming optimization algorithm is specifically as follows: First, an initial population is randomly generated. The position of each search agent represents a candidate solution, and its position vector is randomly generated within a predefined search space, as follows: ; The initial value for each dimension It is the upper and lower bounds given in this dimension. Randomly generated internally The number of dimensions; Then, fitness evaluation is performed, calculating the fitness value of each search agent and recording the global optimal solution in the current population. The energy coefficient for the current iteration is calculated using the energy formula; this coefficient controls the algorithm's transition from exploration to development. The calculation formula is as follows: ; in, This represents the maximum number of iterations. like Perform exploratory behavior: Long-distance migration simulates the large-scale migration of lemming populations. Its location update formula integrates the current optimal solution, random agents, and Brownian motion, as follows: ; in, This is the new position of the search agent in the next iteration t+1. This is the current optimal solution. As a directional marker, Let the vector be a Brownian motion vector. For random control vectors, For the search individual randomly selected from the population; Burrowing simulates the local digging behavior of lemmings, randomly exploring the vicinity of their current location. The location update formula is as follows: ; in, A random number related to the current iteration number. This is the current optimal solution. For the search individual randomly selected from the population; like Execute development activities: Foraging simulates the spiral search behavior of lemmings around food sources, exploring locally around the current optimal solution. Its position update formula is as follows: ; in, The spiral shape function simulates a random search path. It is a random number; Avoiding predators simulates the behavior of lemmings fleeing back to a safe burrow, moving towards the current optimal solution. The position update formula is as follows: ; in, The escape coefficient, This is the Levy flight function, used to simulate the random paths lemmings take when escaping, enhancing the diversity of local search.

[0008] Finally, update the optimal solution, evaluate the fitness value of the new position, and update the global historical optimal solution.

[0009] Furthermore, in one embodiment of the present invention, the artificial lemming algorithm is adaptively improved, specifically as follows: The Circle chaotic mapping is used instead of traditional random initialization, and its mathematical formula is as follows: ; in, For the generated next-generation chaotic sequence values, The value of the chaotic sequence in the current iteration; A quantum rotation gate mechanism is introduced into long-distance migration behavior, and the Brownian motion vector is quantized and perturbed to simulate the quantum tunneling effect. The mathematical formula is as follows: ; in, The original Brownian motion vector, This is the new vector after quantization perturbation.

[0010] A mirror-backward learning strategy is incorporated into foraging behavior. This strategy generates and evaluates mirror solutions to the current solution, and its mathematical formula is as follows: ; in, For the generated mirror solution, This is the current globally optimal solution. This represents the current position of the current agent. An adaptive dimensionality decay factor is embedded in the energy model, and its mathematical formula is shown below: ; in, As an energy factor, The number of dimensions; Through the synergistic effect of these improvement strategies, the algorithm, while maintaining the original framework, exhibits superior global exploration capabilities, local development accuracy, and dimensional adaptability when solving UAV 3D path planning problems.

[0011] Furthermore, in one embodiment of the present invention, the real-time dynamic path replanning specifically includes: During the flight of the UAV, the system monitors the dynamic changes in the environment and the operating status of the UAV in real time, and updates the flight environment model based on the latest perceived environmental data to ensure that it flies stably along the predetermined flight path; If a significant change in the environment is detected or the drone's status deviates from the original planned path, the trajectory will be dynamically adjusted based on real-time information to optimize the path smoothness and ensure the safety and flyability of the path.

[0012] This invention proposes a UAV 3D path planning method based on an improved artificial lemming optimization algorithm, enabling UAVs to navigate autonomously in complex environments, avoid obstacles, find the optimal path, and improve mission efficiency and safety. Specific beneficial effects include: 1. The present invention uses airborne sensors to acquire environmental data and preprocess the environmental data. Combined with mission requirements, a high-fidelity three-dimensional flight environment model is constructed, which provides accurate and reliable input for path planning and lays the foundation for the application of the algorithm. 2. The improved artificial lemming optimization algorithm employed in this invention enhances the diversity and ergodicity of initial solutions by introducing a Circle chaotic mapping for population initialization. During the long-distance migration phase, a perturbation strategy based on a quantum rotating gate is introduced to simulate the quantum tunneling effect, enabling the algorithm to escape local optima traps with a higher probability. In the development phase, a mirror-backward learning mechanism is used to generate and evaluate mirror solutions in the symmetric direction of the current optimal solution, expanding the search range for local regions. An adaptive dimensionality decay factor is introduced into the energy model, allowing the algorithm's random perturbation intensity to adaptively adjust as the problem dimension increases, improving the algorithm's high-dimensional scalability and adaptability. This significantly improves the quality of solutions, global exploration capability, convergence speed, and stability under high-dimensional problems, thereby generating better flight paths. 3. This invention enables UAVs to quickly adjust their paths during flight based on real-time environmental and personal status changes through dynamic monitoring and online replanning mechanisms, ensuring the real-time safety, flyability, and practicality of the generated paths, and enhancing the overall autonomy and reliability of the system. Attached Figure Description

[0013] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of a three-dimensional UAV path planning method based on an improved artificial lemming optimization algorithm, as described in Implementation Method 1. Figure 2 This is a model diagram of the three-dimensional UAV path planning system described in Implementation Method 1; Figure 3 This is a flowchart of the improved artificial lemming optimization algorithm described in Implementation Method 1. Detailed Implementation

[0014] Various embodiments of the present invention will now be clearly and completely described with reference to the accompanying drawings. The embodiments described with reference to the drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0015] Implementation Method 1: This implementation method describes a UAV 3D path planning method based on an improved artificial lemming optimization algorithm, such as... Figure 1 As shown, specifically: A system model is constructed, environmental data is acquired using airborne sensors, and the environmental data is preprocessed to form a three-dimensional flight environment model of the UAV in combination with mission requirements. A problem-solving framework is established, mapping the constraints (obstacle avoidance, terrain constraints) and objective functions (path length, smoothness, etc.) of UAV 3D path planning to the search space and fitness function of the optimization algorithm. The artificial lemming optimization algorithm is used to search for the optimal flight path, and the path is optimized by combining UAV performance constraints, terrain features and meteorological conditions. Algorithm adaptability optimization was carried out. For the 3D path planning problem of UAVs under complex conditions, the artificial lemming algorithm was adaptively improved. The algorithm parameters and structure were adjusted according to the mathematical characteristics of the problem model (such as the continuity of terrain functions and the convexity of obstacles) in the initialization, exploration and development stages. Real-time dynamic path replanning adjusts the flight path based on real-time environmental changes (such as sudden obstacles and weather disturbances) and the drone's own status information perceived by sensors during flight, in order to meet safety and mission requirements.

[0016] Specifically, the construction of the three-dimensional flight environment model includes the following aspects, such as... Figure 2 As shown: Multi-source environmental data is collected using sensors, and combined with the drone's current location and mission requirements, a three-dimensional flight environment model of the drone is constructed, including information such as terrain height and obstacle distribution.

[0017] The starting point of the path planning is defined as The target point is defined as The requirement for UAV 3D path planning is to obtain an optimal route while taking into account all obstacle threats and its own constraints, where obstacle threats are represented in the form of cylinders.

[0018] The specific execution steps are as follows: Let the starting point of the drone flight be... The destination is Based on cubic spline interpolation, a smooth curve is generated, and the objective function is expressed as: ; in, For the total cost, For path length cost, For the cost of a high standard deviation, Cost of planning path smoothness The weights are defined by the following constraints on the weight coefficients: ; Drone flights typically need to save time and reduce costs as much as possible while ensuring safety; therefore, the length of the path planning is crucial. The mathematical model is shown below: ; in, For the first step in planning the path of the drone There are several waypoints. Furthermore, the drone's flight altitude has a significant impact on the control system and safety, as shown in the mathematical model below: ; Finally, the impact of the drone turning also needs to be considered, and the mathematical model is shown below: ; in, for .

[0019] In summary, the model for the UAV path planning optimization problem can be obtained, and the formula is: ; in, For flightable paths, and These are the ground and obstacles, respectively. The ground and obstacle sets are modeled in the study as follows: Furthermore, the artificial lemming optimization algorithm is used to search for the optimal flight path, and the path is optimized by combining the performance constraints of the UAV, terrain features and meteorological conditions. The steps of the artificial lemming optimization algorithm are as follows: The ALA algorithm is a swarm intelligence optimization algorithm that requires an initialization step before iterative optimization begins. The core of this step is establishing the initial positions of the entire population (i.e., the set of all candidate solutions). All candidate solutions form an N x Dim matrix. The rows represent the size of the population, i.e., how many individuals are included in the solution, and the columns represent the number of dimensions of the problem.

[0020] ; ; The initial value for each dimension It is the upper and lower bounds given in this dimension. The population is randomly generated to ensure that the initial population is evenly distributed throughout the possible search space.

[0021] Lemmings undertake random long-distance migrations when food is scarce (usually due to overcrowding), exploring vast search areas in search of habitats with richer food resources. Their location update formula is as follows: ; ; ; ; in, This is a direction indicator. It is a factor that randomly takes the value 1 or -1, used to change the search direction, help the algorithm escape local optima, and increase the chance of a rigorous search within the problem domain.

[0022] This is a Brownian motion vector. It consists of random numbers following a standard normal distribution (mean 0, variance 1), providing a dynamic and uniform step size for exploring potential new regions in the search space.

[0023] This is a random control vector. It is a vector of size 1 x Dim (dimension number), whose elements are random numbers uniformly distributed in the interval [-1, 1]. This vector is used to control the influence weights of the current optimal individual and random individuals on the migration direction, simulating the interaction between individuals during lemming migration.

[0024] When faced with threats or needing to find more resources, lemmings will randomly dig new burrows based on the location of their current burrow and the locations of other random individuals in the population. In the algorithm, this behavior is abstracted as a form of local fine-grained search to avoid missing local optima. The position update formula is as follows: ; ; in, This is the new position of the search agent in the next iteration t+1. The search individual is randomly selected from the population.

[0025] The step size control factor is a dynamic, periodically changing, and random step size coefficient that determines the movement range of the digging behavior.

[0026] Furthermore, artificially brooded lemmings exhibit foraging behavior, a local exploitation strategy. Through spiral paths and random perturbations, it guides the search agent to conduct intensive and meticulous searches near historical optimal solutions, aiming to fully exploit the potential of local areas and increase the probability of finding high-quality solutions. Its position update formula is as follows: ; ; ; in, This is a spiral shape function used to generate spiral paths, simulating the search pattern of lemmings hovering and circling around the optimal solution.

[0027] Let be the radius of the foraging range, which is the Euclidean distance between the current position and the optimal solution.

[0028] During the algorithm's development phase, the artificial lemmings also exhibit predator avoidance behavior. When the lemmings (the search agents in the algorithm) detect a predator (analogous to a local optimum or threat), they do not flee completely randomly, but quickly return to the safety of their burrows (i.e., the currently known global optimum). Simultaneously, during their escape, they employ deceptive maneuvers (such as sudden changes of direction) to confuse and evade their pursuers. Its core objective is to enable the search agent to effectively return and perform a depth search around the current optimum, while introducing random perturbations to avoid premature stagnation. The position update formula is as follows: ; ; ; in, The escape coefficient is a coefficient that decreases linearly with the number of iterations t. In the early stages of the algorithm, A larger value allows individuals to escape and search over a wider range; as the algorithm progresses, The value decreases, making the search increasingly focused on the vicinity of the optimal solution, thus achieving refined development.

[0029] This is the Lévy flight function. It's a random path characterized by occasional long-distance jumps, used here to simulate unpredictable, deceptive maneuvers during escape. It helps the algorithm introduce bursts of exploration in local exploitation, effectively escaping local optima.

[0030] In the Artificial Lemming Algorithm (ALA), an energy factor is used. This allows for a smooth transition from "exploration" to "exploitation," gradually decreasing during the iterative process. When lemmings have sufficient energy, they selectively migrate or burrow; otherwise, they forage and evade predators. The energy factor is calculated using the following formula: ; ; ; in, This constitutes the overall trend of energy decay. With the number of iterations... Approaching the maximum number of iterations The entire value will smoothly decrease from a large number and approach 0, ensuring that the algorithm as a whole can transition from exploration-driven to development-driven.

[0031] The algorithm sets the threshold to 1, when When lemmings are active, they tend to explore (e.g., migrate, dig burrows); when At other times, they turn to development (such as foraging for food or avoiding predators). Throughout the entire iteration process... The probability is about 50%, which ensures that exploration and development activities occur in a balanced way, so that the algorithm can naturally and smoothly transition from global search to local fine search.

[0032] Then, the 3D path planning problem of UAVs is combined with the Artificial Lemming Algorithm (ALA), which is mapped to the search space and fitness function optimized by the ALA algorithm.

[0033] Each lemming represents a potential solution, i.e., a complete flight path. The physical coordinates of the path are defined by the boundary of the search space (LB, UB), used to initialize the population and constrain position updates, ensuring the path remains within the mission airspace. Specifically, it is represented as follows: ; Path cost (length, height, smoothness) is the main component of the fitness function in the algorithm, constituting the primary goal of the algorithm's search development and guiding the algorithm to optimize path quality. The fitness function is expressed as follows: ; Among them, obstacles and terrain constraints are penalty terms in the fitness function. This transforms constraints into penalties, driving the algorithm to perform a global exploration to avoid infeasible regions, specifically as follows: ; in, This is the collision penalty coefficient (maximum value). The number of collisions with the ground. This represents the number of collisions with obstacles.

[0034] The goal of finding the optimal feasible path is to minimize the fitness function. The value of is such that a feasible solution (no collisions) has a penalty term of 0, and its fitness equals the total cost; an infeasible solution will have a very poor fitness due to a large penalty term, and will thus be eliminated by the algorithm. The global optimum is found by balancing exploration (avoiding collisions) and development (optimizing cost).

[0035] Furthermore, for the 3D path planning problem of UAVs under complex conditions, the artificial lemming algorithm is adaptively improved. In the initialization, exploration, and development stages, the algorithm parameters and structure are adjusted according to the mathematical characteristics of the problem model (such as the continuity of terrain functions and the convexity of obstacles). The overall process is as follows: Figure 3 As shown; This implementation first establishes a three-dimensional flight environment model of the UAV based on environmental data acquired by sensors, and then uses Circle chaotic mapping to initialize the population of all search agents. Then, in each iteration of the algorithm, the fitness value of each agent is calculated (to evaluate path length, smoothness, and safety), and the current optimal path is determined. Simultaneously, the energy coefficient E is calculated based on an improved adaptive energy model.

[0036] If E>1, the algorithm enters the exploration phase, where it updates the path by choosing either "long-distance migration" (which incorporates a quantum rotating gate to enhance global exploration capabilities) or "hole-digging" behavior with a certain probability, thus broadly searching the solution space and avoiding premature convergence. If E<1, the algorithm enters the development phase, where it selects either "foraging" (combined with mirror learning to improve local development accuracy) or "avoiding predators" behavior with a certain probability, thus finely optimizing the path near the current optimal path.

[0037] After each iteration, the historical best solution is updated until the maximum number of iterations is reached, ultimately outputting a safe, smooth, and approximately optimal 3D flight path for the UAV. The entire process, through a structured exploration-development mechanism and multiple improvement strategies, ensures the efficiency, stability, and reliability of path planning.

[0038] The specific scheme for adaptively improving the artificial lemming algorithm is as follows: In the ALA algorithm, random initialization may lead to uneven population distribution, missing potentially high-quality solutions. This implementation uses the Circle chaotic mapping method to generate uniformly distributed initial solutions, covering the entire search space and expanding the search range. The mathematical formula is as follows: ; The original ALA algorithm's long-distance migration behavior employs Brownian motion, limiting its search to the current random direction. This weak directionality can lead to blind oscillations within the search space, resulting in low convergence efficiency, especially in complex multimodal problems. This implementation introduces a quantum rotation gate, superimposing a new, orthogonal random component onto the existing random direction. This is equivalent to searching in a two-dimensional (or higher-dimensional) plane, rather than the classic one-dimensional random walk. This expansion of the search dimension directly enhances exploration capabilities and the ability to escape local optima. The mathematical formula is as follows: ; This new orthogonal component provides the ability to "leap sideways" away from the current search trend. When an algorithm gets stuck in a local optimum, classical Brownian motion may only linger at the bottom. However, the lateral force generated by the quantum rotating door has the probability of directly pushing the solution over the barrier of the valley, achieving a "tunneling" effect and reaching another potentially better region.

[0039] During the foraging phase of the algorithm, the original ALA algorithm relies on a spiral function to generate candidate solutions. While the spiral path can effectively search around the current optimal solution, its search trajectory is deterministic or strongly patterned. This mechanism may lead to two problems: search blind spots and getting stuck in local concavities. This invention introduces a mirror-backward learning strategy during the development phase. Through a mathematical mirror transformation, the search range is expanded from one side to both sides, constructing a symmetric point of the current candidate solution with respect to the current optimal solution. The mathematical formula is as follows: ; This strategy addresses the structural defects of the spiral search operator by introducing a mirror back-learning mechanism to construct a symmetric and competitive local search framework. This significantly improves the search efficiency, solution accuracy, and robustness of the algorithm during the development phase, enabling it to more reliably approximate the global optimum when solving UAV 3D path planning problems.

[0040] In the ALA algorithm, the energy factor The search space is a crucial pivot point for controlling the transition of the algorithm from exploration to development. However, in high-dimensional optimization problems, the volume of the search space grows exponentially, severely weakening the dimensionality scalability of the algorithm, resulting in premature convergence and a significant decrease in solution quality. This invention introduces a dimension-based adaptive decay factor into the energy formula, the mathematical formula of which is shown below: ; In high-dimensional problems, the newly added factor automatically reduces the perturbation intensity, preventing the algorithm from engaging in inefficient and excessive random exploration in a vast search space, and significantly improving the algorithm's dimensionality scalability across different dimensional problems. Furthermore, this balance is dynamically recalibrated to ensure that in high-dimensional environments, the algorithm maintains the necessary exploration capabilities to avoid premature convergence while concentrating more resources on promising regions for development.

[0041] An improved artificial lemming optimization algorithm was used to solve the problem, and the flight path of the UAV was obtained by combining the performance constraints of the UAV, terrain features and weather conditions.

[0042] Furthermore, during flight, the system monitors environmental changes in real time, adjusts the flight path accordingly based on these changes and the UAV's status information, and iteratively calculates and outputs the optimal solution. Specifically: During flight, the system monitors dynamic changes in the environment and the operational status of the UAV in real time. In each algorithm iteration, the flight environment model is updated based on the latest perceived environmental data, while continuously acquiring state parameters such as the UAV's attitude, battery level, position, and speed to ensure stable flight along the predetermined path.

[0043] If a significant change in the environment is detected or the drone's status deviates from the original planned path, the trajectory will be dynamically adjusted based on real-time information to optimize path smoothness and perform online replanning to ensure path safety and flyability.

[0044] The above provides a detailed description of a UAV 3D path planning method based on an improved artificial lemming optimization algorithm proposed in this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for UAV 3D path planning based on an improved artificial lemming optimization algorithm, characterized in that, Specifically: A system model is constructed, environmental data is acquired using airborne sensors, and the environmental data is preprocessed to form a three-dimensional flight environment model of the UAV in combination with mission requirements. A problem-solving framework is established, mapping the constraints and objective function of UAV 3D path planning to the search space and fitness function of the optimization algorithm. The artificial lemming optimization algorithm is used to search for the optimal flight path, and the path is optimized by combining UAV performance constraints, terrain features and meteorological conditions. Algorithm adaptability optimization was carried out. For the three-dimensional path planning problem of UAVs under complex conditions, the artificial lemming algorithm was adaptively improved. The algorithm parameters and structure were adjusted according to the mathematical characteristics of the problem model in the initialization, exploration and development stages. Real-time dynamic path replanning adjusts the flight path based on real-time environmental changes and the drone's own status information perceived by sensors during flight, in order to meet safety and mission requirements.

2. The UAV 3D path planning method based on an improved artificial lemming optimization algorithm according to claim 1, characterized in that, The three-dimensional flight environment model is specifically as follows: Multi-source environmental data is collected using sensors, and combined with the UAV's current location and mission requirements, a 3D flight environment model of the UAV is constructed. The objective function is expressed as: ； in, As weight, For path length cost, For the cost of a high standard deviation, Cost of planning path smoothness For the total cost; The model of the ground and obstacles is represented as follows: ； in, and These are the horizontal and vertical coordinates in three-dimensional space, used to locate the drone's position on the horizontal plane. The vertical height of the terrain and obstacles is used to mathematically simulate complex 3D terrain and obstacle distributions to detect whether the path intersects with the ground or obstacles.

3. The UAV 3D path planning method based on an improved artificial lemming optimization algorithm according to claim 1, characterized in that, The artificial lemming optimization algorithm is as follows: First, an initial population is randomly generated. The position of each search agent represents a candidate solution, and its position vector is randomly generated within a predefined search space, as follows: ； The initial value for each dimension It is the upper and lower bounds given in this dimension. Randomly generated internally The number of dimensions; Then, fitness evaluation is performed, calculating the fitness value of each search agent and recording the global optimal solution in the current population. The energy coefficient for the current iteration is calculated using the energy formula; this coefficient controls the algorithm's transition from exploration to development. The calculation formula is as follows: ； in, This represents the maximum number of iterations. like Perform exploratory behavior: Long-distance migration simulates the large-scale migration of lemming populations. Its location update formula integrates the current optimal solution, random agents, and Brownian motion, as follows: ； in, This is the new position of the search agent in the next iteration t+1. This is the current optimal solution. As a directional marker, Let the vector be a Brownian motion vector. For random control vectors, For the search individual randomly selected from the population; Burrowing simulates the local digging behavior of lemmings, randomly exploring the vicinity of their current location. The location update formula is as follows: ； in, A random number related to the current iteration number. This is the current optimal solution. For the search individual randomly selected from the population; like Execute development activities: Foraging simulates the spiral search behavior of lemmings around food sources, exploring locally around the current optimal solution. Its position update formula is as follows: ； in, The spiral shape function simulates a random search path. It is a random number; Avoiding predators simulates the behavior of lemmings fleeing back to a safe burrow, moving towards the current optimal solution. The position update formula is as follows: ； in, The escape coefficient, The Lévy flight function is used to simulate the random paths taken by lemmings when escaping, enhancing the diversity of the local search. Finally, the optimal solution is updated, the fitness value of the new position is evaluated, and the global historical optimal solution is updated.

4. The UAV 3D path planning method based on the improved artificial lemming optimization algorithm according to claim 1, characterized in that, The artificial lemming algorithm has been adaptively improved, specifically as follows: The Circle chaotic mapping is used instead of traditional random initialization, and its mathematical formula is as follows: ； in, For the generated next-generation chaotic sequence values, The value of the chaotic sequence in the current iteration; A quantum rotation gate mechanism is introduced into long-distance migration behavior, and the Brownian motion vector is quantized and perturbed to simulate the quantum tunneling effect. The mathematical formula is as follows: ； in, The original Brownian motion vector, The new vector after quantized perturbation; A mirror-backward learning strategy is incorporated into foraging behavior. This strategy generates and evaluates mirror solutions to the current solution, and its mathematical formula is as follows: in, For the generated mirror solution, This is the current globally optimal solution. This represents the current position of the current agent. An adaptive dimensionality decay factor is embedded in the energy model, and its mathematical formula is shown below: ； in, As an energy factor, The number of dimensions; Through the synergistic effect of these improvement strategies, the algorithm, while maintaining the original framework, exhibits superior global exploration capabilities, local development accuracy, and dimensional adaptability when solving UAV 3D path planning problems.

5. The UAV 3D path planning method based on an improved artificial lemming optimization algorithm according to claim 1, characterized in that, The real-time dynamic path replanning specifically includes: During the flight of the UAV, the system monitors the dynamic changes in the environment and the operating status of the UAV in real time, and updates the flight environment model based on the latest perceived environmental data to ensure that it flies stably along the predetermined flight path; If a significant change in the environment is detected or the drone's status deviates from the original planned path, the trajectory will be dynamically adjusted based on real-time information to optimize the path smoothness and ensure the safety and flyability of the path.

Images