Unmanned surface vehicle path planning method based on improved cuckoo algorithm

Abstract

The invention discloses an unmanned surface vehicle path planning method based on an improved cuckoo algorithm. According to the method, a self-adaptive step length updating mechanism combining a pheromone factor and a heuristic factor is introduced in the iteration process of the cuckoo algorithm. The pheromone factor refers to a positive feedback mechanism of an ant colony algorithm to guide search to be carried out towards a historical high-quality solution direction; the heuristic factor dynamically adjusts the step length direction according to the distance between the current optimal solution and the candidate solution, and the local development capability is enhanced; and through a dynamic weight adjustment strategy, self-adaptively balancing global exploration and local search in an iteration process, and performing connectivity optimization on discrete path points output by an algorithm by using a Dijkstra algorithm to obtain a continuous and collision-free sailing path. According to the method, the convergence speed and the solving quality of path planning are remarkably improved, and the autonomous navigation capability of the unmanned surface vehicle in a complex environment is enhanced.

Description

Technical Field

[0001] This invention belongs to the technical field of ship path planning, specifically relating to a path planning method for unmanned surface vessels based on an improved cuckoo algorithm. Background Technology

[0002] Compared to traditional manned vessels, unmanned surface vessels (USVs) offer advantages such as smaller size, higher maneuverability, and controllable risks, enabling them to perform complex tasks in harsh and high-risk marine environments. They are widely used in marine surveying, island and reef patrols, maritime traffic monitoring, military reconnaissance, and emergency rescue, playing an increasingly crucial role in modern marine engineering. The path planning method directly determines the safety of USV navigation and the reliability of mission completion.

[0003] Path planning algorithms can currently be divided into two categories: traditional algorithms and intelligent algorithms. Traditional algorithms can be further divided into geometric model search algorithms, artificial potential field algorithms, and probabilistic abstraction algorithms. Intelligent algorithms can be divided into swarm intelligence algorithms, evolutionary algorithms, and human-heuristic algorithms. Traditional algorithms have poor adaptability to changes in the environmental space. Human-heuristic methods, such as reinforcement learning and deep learning, while showing potential in handling complex environments and achieving autonomous decision-making, are limited in their application in resource-constrained underwater environments due to their high resource consumption, long training time, and reliance on large amounts of sample data. In contrast, swarm intelligence algorithms exhibit unique advantages, possessing strong adaptability, robustness, and parallel computing capabilities, while having relatively low resource requirements.

[0004] Compared to similar algorithms, the Cuckoo algorithm has advantages such as fewer parameters and simpler implementation. Its synergy with the Lévy flight mechanism endows the algorithm with excellent global optimization capabilities.

[0005] The implementation of the cuckoo algorithm will be based on the following three idealized rules: Rule 1: Each cuckoo lays only one egg at a specific time and randomly selects a parasitic nest to place the egg.

[0006] Rule 2: In each generation, the parasitic nest with the best hatching conditions among all randomly selected parasitic nests is reserved for the next generation, which is the elite retention mechanism.

[0007] Rule 3: The number of selectable parasitic nests is fixed. The host bird has a certain probability of discovering the foreign bird's egg, and the probability of discovery is [not specified]. Once discovered, the host bird has two choices: either destroy the egg or push it out of the nest, or abandon the nest and choose a new location to build a new nest.

[0008] In 2009, Yang and Deb proposed the Cuckoo Search Algorithm, which effectively solves complex optimization problems by simulating the parasitic breeding behavior of cuckoos. Li X et al. proposed a new update strategy based on optimal individuals and random individuals, combining the new strategies through a linearly decreasing probability rule. They also proposed a new search strategy based on orthogonal learning to enhance the search capability of the basic Cuckoo Search Algorithm.

[0009] Liu Qing et al. proposed a multi-strategy fusion golden sine cuckoo search algorithm, introducing the Latin hypercube sampling method to initialize the population and enhance the population diversity of the CS algorithm. Xie Yongsheng et al. incorporated genetic operators, 2-opt, the Metropolis criterion of simulated annealing, and insertion, exchange, and reversal methods into the cuckoo search algorithm.

[0010] In the path planning task in complex environment, the traditional cuckoo algorithm has the following shortcomings: (1) The Levi flight search process is too random and lacks an effective guidance mechanism, resulting in a large degree of randomness in the search direction; (2) The convergence speed is slow; (3) It is easy to get trapped in local optima in complex environment, which affects the reliability of path planning. Summary of the Invention

[0011] In view of this, the present invention provides a path planning method for unmanned surface vessels based on an improved cuckoo algorithm, which solves the problems of strong randomness in search direction, slow convergence speed and easy getting trapped in local optima in complex marine environments by the traditional cuckoo algorithm, thereby improving the reliability, safety and mission execution efficiency of unmanned surface vessel path planning.

[0012] The technical solution for implementing the present invention is as follows: A path planning method for unmanned surface vessels based on an improved cuckoo algorithm includes the following steps: Step 1: Establish a two-dimensional rasterized environment model and set the starting and ending points of the unmanned surface vessel; Step 2: Initialize the cuckoo population, maximum number of iterations, and set the pheromone matrix and related parameters; Step 3: During the iteration of the Cuckoo algorithm, an adaptive step size update mechanism that combines pheromone factors and heuristic factors and dynamically adjusts the weights is introduced to generate a new solution based on the improved Lévy flight formula. Step 4: Calculate the fitness of the new solution and retain the better individual as a new candidate solution; Step 5: Based on the discovery probability, discard some solutions, generate new nests, and apply boundary constraints; Step 6: Update the pheromone factor and evaluate historical solutions to avoid premature convergence; Step 7: Recalculate the fitness values ​​of all solutions and retain the better solutions; Step 8: Repeat the iteration until the termination condition is met, and output the set of optimal path points; Step 9: Use Dijkstra's algorithm to perform connectivity processing on the optimal path point set to obtain a complete and continuous collision-free path.

[0013] Furthermore, the adaptive step size update mechanism in step 3 includes: The pheromone factor, whose value is updated based on the performance of historical solutions; The heuristic factor is calculated based on the distance between the current solution and the global optimum. A dynamic weight adjustment strategy is used to adjust the weights of the pheromone factor and the heuristic factor during the iteration process.

[0014] Furthermore, the update formula for the pheromone factor is:

[0015] in, Indicates the pheromone evaporation rate; It is to evaluate candidate solutions The objective function value for quality is called the individual fitness function, which is typically used in minimization problems. The smaller the value, the better the candidate solution; Let denot be the pheromone of the i-th individual in the t-th iteration.

[0016] Furthermore, the heuristic factor The calculation formula is:

[0017] in, This is the current solution; This is the current optimal solution; Is the current solution With the global optimal solution The Euclidean distance between them reflects the degree of similarity or proximity between them.

[0018] Furthermore, in the dynamic weight adjustment strategy, the pheromone weight... and heuristic weights According to the iteration progress The calculation formula is dynamically determined as follows:

[0019] Where t is the current iteration number, It represents the maximum number of iterations.

[0020] Furthermore, the pheromone factor update in step 6 also includes normalizing the pheromone factor:

[0021] in, The normalized pheromone factor, and These represent the minimum and maximum pheromone values ​​for all individuals in the current population, respectively. It is a very small value used to avoid division by zero errors.

[0022] Furthermore, the improved Lévy flight formula in step 3 is as follows:

[0023] in, Step size factor For Lévy random numbers, This is the normalized heuristic factor.

[0024] Furthermore, in step 9, the Dijkstra algorithm is used to process the connectivity of the optimal path point set, specifically by connecting and smoothing discrete path points to ensure the continuity and collision-free nature of the output path.

[0025] Beneficial effects: 1. The method of this invention introduces a pheromone guidance mechanism and a heuristic factor based on the current optimal solution, enabling the algorithm to obtain effective directional guidance in the early stages of iteration, significantly reducing the randomness of the search. The improved algorithm can approach the high-quality solution region more quickly and maintain stable convergence in the later stages, thereby effectively shortening the overall time consumption of path planning.

[0026] 2. This invention creatively borrows the pheromone positive feedback mechanism from the ant colony algorithm to construct an adaptive step size update strategy. Historical high-quality solutions accumulate pheromones to create a guiding effect, making the search process more inclined to proceed towards potential regions, thereby significantly improving the algorithm's global exploration capability in complex environments.

[0027] 3. The method of this invention employs a dynamic weight adjustment strategy to allow pheromone and heuristic factors to play a dominant role at different stages of iteration. In the early stages, pheromone-driven broad exploration is employed, while in the later stages, heuristics drive refined development. This mechanism enables the algorithm to autonomously coordinate "exploration" and "development" behaviors, accelerating convergence to the optimal solution while maintaining population diversity.

[0028] 4. Based on the intelligent algorithm for optimizing and generating discrete path points, the method of this invention further introduces Dijkstra's algorithm from graph theory for connectivity verification and path smoothing. This post-processing step ensures that the final output path is not only theoretically optimal, but also continuous, collision-free, and conforms to the motion constraints of the unmanned surface vessel, and can be directly used for actual navigation control. Attached Figure Description

[0029] Figure 1 The flowchart of the improved cuckoo algorithm of this invention is shown.

[0030] Figure 2 A comparison of the convergence curves of the Cuckoo algorithm before and after the improvement. Detailed Implementation

[0031] This invention provides a path planning method for unmanned surface vessels based on an improved cuckoo algorithm, the specific implementation steps of which are as follows: Step 1. Establish a two-dimensional rasterized environment model and set the starting and ending points of the unmanned surface vessel; Step 2. Initialize the cuckoo population, maximum number of iterations, and set the pheromone matrix and related parameters; Step 3. In the iterative process of the Cuckoo algorithm, an adaptive step size update mechanism that combines pheromone factors and heuristic factors and dynamically adjusts the weights is introduced to generate a new solution based on the improved Lévy flight formula; Step 4. Calculate the fitness of the new solutions and retain the better individuals as new candidate solutions; Step 5. Based on the discovery probability, discard some solutions, generate new nests, and apply boundary constraints; Step 6. Update the pheromone factor and evaluate historical solutions to avoid premature convergence; Step 7. Recalculate the fitness values ​​for all solutions and retain the better solutions; Step 8. Repeat the iteration until the termination condition is met, and output the set of optimal path points; Step 9. Use Dijkstra's algorithm to solve the connectivity problem of the discrete set of optimal path points, and obtain a complete and continuous path that truly satisfies the collision-free requirement.

[0032] The following is in conjunction with the appendix Figure 1 The present invention will be described in detail as follows: 1) Pheromone Update Mechanism. Drawing inspiration from the role of pheromone in path selection in ant colony optimization, this mechanism integrates pheromone factors into the candidate solution update process, guiding the search direction through pheromone concentration. Historically performing solutions retain more pheromone, thus guiding the generation of new solutions.

[0033] There is a pheromone matrix The pheromone level of each individual in a given iteration is denoted as... In each iteration, the pheromone is updated based on the pheromone evaporation rate and individual fitness, expressed as:

[0034] in, Indicates the pheromone evaporation rate; It is to evaluate candidate solutions The objective function value for quality is called the individual fitness function, which is typically used in minimization problems. The smaller the value, the better the candidate solution.

[0035] High adaptability (i.e.) Smaller candidate solutions will attract more pheromones, which helps to attract more searches in subsequent iterations.

[0036] 2) Heuristic Factor. The heuristic factor is calculated based on the distance between the current solution and the optimal solution, encouraging individual solutions to cluster near the optimal solution and improving local search efficiency. Its expression is:

[0037] in, This is the current solution; This is the current optimal solution; It is a candidate solution With the global optimal solution The Euclidean distance between them reflects the degree of similarity or proximity between them. When the candidate solution is closer to the current optimal solution The larger the value, the more valuable the area is for local searching; as the distance increases... The value will gradually decrease, indicating that the solution is relatively less important in the local search.

[0038] 3) Dynamic weight adjustment strategy. In the early stages, it mainly relies on pheromone guidance to broadly search the global space, while in the later stages it gradually shifts to heuristic factor-driven approach, focusing on fine-grained optimization in local regions.

[0039] Let the current iteration progress be:

[0040] in, This represents the maximum number of iterations. The pheromone weights and heuristic weights are defined as follows:

[0041] 4) Normalization of pheromones and heuristic factors. Pheromones may vary significantly across individuals. To ensure comparability of pheromones across individuals, they are normalized, which can be expressed as:

[0042] in, and These represent the minimum and maximum pheromone values ​​for all individuals in the current population, respectively. It is a very small value used to avoid division by zero errors; Normalized The value range is approximately between 0 and 1, which facilitates subsequent combination with heuristic factors.

[0043] To ensure that the heuristic factor and the normalized pheromone have similar dimensions and value ranges, the heuristic factor is also normalized, which can be expressed as:

[0044] in, This represents the maximum value among all heuristic factors for all individuals in the current population, used to classify them as individuals. Scale to the range of 0 to 1.

[0045] This ensures that the two factors have similar scales when merging, preventing one factor from dominating the entire update process due to its excessively large or small scale.

[0046] 5) Comprehensive Influence Factors and New Solution Generation. Combining pheromones and heuristic factors, it can be expressed as:

[0047] Then, combining this with Levi's flight, the step size adjustment is calculated:

[0048] Finally, the formula for generating the new solution (5-9) changes to:

[0049] This update mechanism allows the algorithm to dynamically switch between global and local modes, ensuring both the diversity of the search and accelerating the convergence speed.

[0050] This invention uses the improved Cuckoo algorithm before and after the simulation experiment under the same environment. The comparison of the simulation convergence curves is attached. Figure 2 .

[0051] Analysis of the convergence curves reveals that while the improved algorithm initially has a higher objective function value, it rapidly decreases in the early stages. This indicates that the pheromone factor enables the algorithm to find a better solution early on, while the original algorithm's convergence process is slower, requiring many iterations to break through local optima. In the later stages of the search, the improved algorithm stabilizes earlier, demonstrating that the introduction of the heuristic factor successfully utilizes information from the current global optimum, dynamically adjusting the step size and effectively enhancing local exploitation capabilities, thus accelerating the overall convergence of the algorithm. This demonstrates that the improved strategy designed in this invention effectively improves the convergence speed of the Cuckoo Algorithm.

[0052] In summary, the above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A path planning method for unmanned surface vessels based on an improved cuckoo algorithm, characterized in that, Includes the following steps: Step 1: Establish a two-dimensional rasterized environment model and set the starting and ending points of the unmanned surface vessel; Step 2: Initialize the cuckoo population, maximum number of iterations, and set the pheromone matrix and related parameters; Step 3: During the iteration of the Cuckoo algorithm, an adaptive step size update mechanism that combines pheromone factors and heuristic factors and dynamically adjusts the weights is introduced to generate a new solution based on the improved Lévy flight formula. Step 4: Calculate the fitness of the new solution and retain the better individual as a new candidate solution; Step 5: Based on the discovery probability, discard some solutions, generate new nests, and apply boundary constraints; Step 6: Update the pheromone factor and evaluate historical solutions to avoid premature convergence; Step 7: Recalculate the fitness values ​​of all solutions and retain the better solutions; Step 8: Repeat the iteration until the termination condition is met, and output the set of optimal path points; Step 9: Use Dijkstra's algorithm to perform connectivity processing on the optimal path point set to obtain a complete and continuous collision-free path.

2. The method according to claim 1, characterized in that, The adaptive step size update mechanism in step 3 includes: The pheromone factor, whose value is updated based on the performance of historical solutions; The heuristic factor is calculated based on the distance between the current solution and the global optimum. A dynamic weight adjustment strategy is used to adjust the weights of the pheromone factor and the heuristic factor during the iteration process.

3. The method according to claim 2, characterized in that, The update formula for the pheromone factor is: in, Indicates the pheromone evaporation rate; It is to evaluate candidate solutions The objective function value for quality is called the individual fitness function, which is typically used in minimization problems. The smaller the value, the better the candidate solution; For the first i The first individual t Pheromones are recorded in the next iteration.

4. The method according to claim 2, characterized in that, The heuristic factor The calculation formula is: in, This is the current solution; This is the current optimal solution; Is the current solution With the global optimal solution The Euclidean distance between them reflects the degree of similarity or proximity between them.

5. The method according to claim 2, characterized in that, In the dynamic weight adjustment strategy, the pheromone weight... and heuristic weights According to the iteration progress The calculation formula is dynamically determined as follows: Where t is the current iteration number, It represents the maximum number of iterations.

6. The method according to claim 1, characterized in that, The pheromone factor update in step 6 also includes normalization processing of the pheromone factor: in, The normalized pheromone factor, and These represent the minimum and maximum pheromone values ​​for all individuals in the current population, respectively. It is a very small value used to avoid division by zero errors.

7. The method according to claim 1, characterized in that, The improved Levy flight formula in step 3 is as follows: in, Step size factor For Lévy random numbers, This is the normalized heuristic factor.

8. The method according to any one of claims 1-7, characterized in that, In step 9, Dijkstra's algorithm is used to process the connectivity of the optimal path point set. Specifically, discrete path points are connected and smoothed to ensure the continuity and collision-free nature of the output path.

Images