AUV (Autonomous Underwater Vehicle) path planning method based on multi-strategy collaborative enhanced artificial mouse algorithm

Abstract

The invention discloses an AUV (Autonomous Underwater Vehicle) path planning method based on a multi-strategy collaborative enhanced artificial mouse algorithm, and belongs to the technical field of AUV path planning. In order to solve the problems of slow convergence speed and low convergence precision of an existing path planning algorithm in the field of AUV path planning, the method comprises the following steps: carrying out path normalization by adopting an artificial travel mouse algorithm to obtain an optimal path, taking a connecting line of a starting point and an end point as a center line during initialization, adding random disturbance to a center line interpolation point to generate a path point, and carrying out path planning on the path point; generating an initial population; the path with the smaller fitness is selected as a superior path, and global exploration, hole digging or foraging / pursuer avoiding is decided according to the value of the self-adaptive nonlinear attenuation dynamic energy factor; evaluating the fitness of the individual, and if the fitness of the iteration is more excellent than the optimal fitness of the individual, updating the optimal fitness of the individual; if the fitness is more excellent than the global optimum fitness, updating the global optimum; and finally obtaining a final path based on the global optimal solution.

Description

TECHNICAL FIELD

[0001] The application relates to an AUV path planning method. BACKGROUND

[0002] Path planning technology is an important guarantee for the safe navigation of an underwater vehicle (AUV) in water, and an efficient and stable path planning algorithm is a key to the autonomous navigation of the AUV. When facing the autonomous navigation task of the underwater vehicle in a complex three-dimensional environment, most path planning algorithms have certain limitations. The rapid exploration random tree (RRT) algorithm is adapted to an unknown environment by virtue of a random sampling mechanism, but the path is often not smooth and is non-optimal; the A* algorithm can guarantee completeness in a structured grid map, but the search efficiency thereof is seriously dependent on the environment resolution, and the memory consumption thereof increases sharply with the expansion of the map size; the artificial potential field method is fast in response, but is prone to local optimization. In recent years, heuristic algorithms have been widely applied to the field of path planning of the AUV, and many researches attempt to introduce new meta-heuristic algorithms such as the sparrow search algorithm (SSA), the whale optimization algorithm (WOA) and the Harris hawk optimization (HHO), or to improve traditional swarm intelligence algorithms such as the ant colony optimization (ACO) and the grey wolf optimizer (GWO), so as to improve the convergence performance and path quality by introducing adaptive weights, bidirectional search or hybrid strategies. However, these methods are still generally difficult to balance the convergence speed, path smoothness and global optimality in a real three-dimensional complex scene, and are often restricted in practical application due to parameter sensitivity or high computational complexity. Therefore, it is still a challenging task to realize efficient autonomous navigation in a complex three-dimensional underwater space. SUMMARY

[0003] The application is to solve the problems of slow convergence speed, low convergence accuracy, significant decline in planning ability in a three-dimensional complex environment and difficulty in planning a safe and reliable path with the minimum cost of the existing path planning algorithm in the field of AUV path planning.

[0004] An AUV path planning method based on a multi-strategy cooperative enhancement type artificial vole algorithm, comprising the following steps: discretizing the path into j intermediate control points , obtaining the path by adding the start point S and the end point G, and representing each candidate path individual as , adopting the artificial vole algorithm to perform path optimization to obtain an optimal path; The process of adopting the artificial vole algorithm to perform path optimization comprises: adopting a directional initialization strategy to initialize in an initialization stage: taking the center line of the start point-end point as the center line, generating path points by adding random disturbances at the center line interpolation points, and then generating an initial population; selecting a path with a smaller fitness as a superior path, calculating an adaptive nonlinear decay dynamic energy factor , a weight factor for controlling the influence of diversity, a nonlinear factor for controlling the time decay; a maximum number of iterations; a normalized initial diversity; according to the value of the dynamic factor E(t), global exploration is performed, or hole digging is performed, or foraging / evading predators is performed; Then boundary and obstacle checking is performed; the fitness of the current individual is evaluated, if the fitness of this iteration is better than the optimal fitness of the current individual, the individual optimal is updated; if the fitness is better than the global optimal fitness, the global optimal is updated; then it is judged whether the maximum number of iterations is reached, if not, the next iteration is entered; if so, the global optimal solution is output; and the final path is obtained based on the global optimal solution.

[0005] Further, a random disturbance is added to the center line interpolation point to generate a path point: , represents the jth path point of the ith individual; represents the jth path point of the basic individual; perturbation represents a random disturbance value.

[0006] Further, the random disturbance value is U, which is a matrix with a size of 1 3, where each element is a uniformly distributed random number in [0, 1]; N is the population size, i.e., the number of individuals in the population; UB is the upper bound of the three-dimensional coordinate, and LB is the lower bound of the three-dimensional coordinate.

[0007] Further, the smaller the fitness, the better the path fitness function in the process of selecting the superior path is

[0008] wherein, is the path length, is the turning angle, is the obstacle avoidance property; is the weight of each index, which is set to in the present embodiment and subsequent experiments. represents the maximum path length, represents the maximum turning cost, represents the maximum obstacle avoidance risk.

[0009] Further, the obstacle avoidance property is as follows: ,

[0010] i.e., infinity, a large value is given according to the actual situation.

[0011] Further, the normalized initial diversity where, to prevent the extremely small value except 0; N is the population size; Dim is the dimension of three-dimensional space; represents the average value of all individuals in the kth dimension at the iteration number t; represents the value of the kth dimension of the ith individual at the iteration number t; is the population center, which is the average value vector of all individuals in the current population in each dimension.

[0012] Further, the process of performing global exploration, or performing hole digging, or performing foraging / evading predator according to the value of the dynamic factor E(t) includes: a. If the energy factor E(t) is greater than or equal to the first threshold value, global exploration is performed, Brown motion simulation of irregular, random migration path is used by introducing adaptive Brown step length, and updating is performed by combining the difference between the current optimal solution and the random individual; In the process of performing global exploration using Brown motion simulation of irregular, random migration path by introducing adaptive Brown step length where, is the current individual path; is the current optimal path; is a randomly selected individual in the population; is a direction control vector; is a parameter for changing the search direction; is a normal variable of Brown motion, where, , is the maximum step length; is the minimum step length; b. If the first threshold value < E(t) < second threshold value, a mixed strategy is adopted, hole digging is performed, and the population center is introduced in the hole digging mechanism, regional exploration is performed around the population center, i.e. where, : fluctuation is created over time; rand is a random number between [0, 1); F is as above; is a random individual; ; c. If E(t) is less than or equal to the second threshold value, foraging / evading predator is performed.

[0013] Further, the process of foraging / evading predator is as follows: Foraging: fine-tuning is performed by using a spiral mechanism in combination with the positional relationship between the current individual and the optimal solution; ,

[0014] where, Euclidean distance between the optimal individual and the current individual; Evasion of pursuer: simulate escape behavior through Lévy flight mechanism, form "long jump" strategy, and avoid falling into local optimum;

[0015] wherein, Adopt square root attenuation mode, gradually reduce with iteration.

[0016] Further, the wherein, is the initial jump radius; t is the current iteration number; is the maximum iteration number.

[0017] Further, in the process of obtaining the final path based on the global optimal solution, a natural cubic spline interpolation method is adopted to perform smoothing operation on the path corresponding to the global optimal solution to obtain the final path. Beneficial effects

[0018] Based on the multi-strategy synergistically enhanced artificial vole algorithm, firstly, a directional initialization strategy is introduced to accelerate the convergence speed of the algorithm and ensure the real-time performance of planning. Secondly, the adaptive nonlinear decay balance factor is adopted for the algorithm itself to balance exploration and development, and the population diversity is added as a judgment factor in the behavior selection of exploration and development, so that the convergence stability is improved; the adaptive Brown step length is introduced in the global exploration stage, so that the search ability is improved in the early exploration stage, and the local optimum is avoided; a controlled Lévy flight mechanism is proposed in the local development stage to replace the original Lévy flight, so that the jump radius is dynamically limited, and the excellent solution is avoided in the later stage; in addition, the algorithm adds a hybrid strategy stage, which balances the global and local directions around the population center; finally, the natural spline interpolation method is adopted to smooth the final path, so that it can meet the actual motion state of the AUV, and the application has excellent planning ability in the three-dimensional complex environment. The simulation and experimental results show that the multi-strategy synergistically enhanced artificial vole algorithm can quickly generate a safe and reliable global path with the minimum cost in the three-dimensional underwater environment, and can be used for path planning of the AUV. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 It is a path planning process schematic diagram of the application.

[0020] Figures 2-4 It is a simulation diagram of different perspectives with the starting point being [10 50 1] and the end point being [350 240 100].

[0021] Figures 5-7Simulation figure of different perspectives with start point [150 100 1] and end point [350 400 100].

[0022] Figures 8-9 Simulation figure of different perspectives with start point [10 50 1] and end point [350 240 100] (after path smoothing effect).

[0023] Figures 10-11 Simulation figure of different perspectives with start point [150 100 1] and end point [350 400 100] (after path smoothing effect).

[0024] Figures 12-14 Simulation figure of different perspectives of path planning in real sea bottom environment with start point [50, -300, -1000] and end point [-368, 432, -4000].

[0025] Figures 15-17 Simulation figure of different perspectives of path planning in real sea bottom environment with start point [-349, -410, -1000] and end point [-203, 490, -4000]. DETAILED DESCRIPTION DETAILED EMBODIMENT I:

[0027] As shown in Figure 1 , the embodiment is an AUV path planning method based on a multi-strategy collaborative enhanced artificial traveling mouse algorithm, comprising the following steps: (1) Model the environment to which the AUV belongs, and determine the coordinate positions of the start point and the end point;

[0028] (2) Introduce a directional initialization strategy for population initialization: In the initialization process, the initial path set is obtained, so the path is discretized into j intermediate control points , plus the start point S and the end point G:

[0029] The optimization is , so each candidate path individual (traveling mouse) can be represented as:

[0030] In the initialization stage, a directional initialization strategy is introduced to accelerate the convergence speed of the algorithm. Referring to the actual underwater obstacle environment, most of the optimal paths are distributed around the connecting line, and the algorithm generates the initial population by taking the connecting line of the start point and the end point as the center line, adding random disturbance to the center line interpolation points during population initialization.

[0031]

[0032] where, U is a 1 3 matrix, where each element is a uniformly distributed random number in [0, 1]. N is the population size, i.e., the number of individuals in the population. UB is the upper bound of the three-dimensional coordinate, LB is the lower bound of the three-dimensional coordinate; perturbation represents a random perturbation value, which is randomly generated around the ideal optimal solution while ensuring exploration; The path points are generated by adding random perturbations to the center line interpolation points:

[0033] where, represents the jth path point of the ith individual. represents the jth path point of the base individual, which is the straight line path connecting the start point and the end point, and the random center interpolation point in the path is .

[0034] Calculate the path fitness function

[0035] where, is the path length, is the turning angle, is the obstacle avoidance property; is the weight of each index, which is set to in this embodiment and subsequent experiments. represents the maximum path length, represents the maximum turning cost, represents the maximum obstacle avoidance risk, , , is used to normalize the reference value.

[0036] The path length is as follows:

[0037] where .

[0038] The turning angle (curvature constraint) is as follows: ,

[0039] where, is the turning angle of the path at point , i.e., the turning angle between two adjacent path segments.

[0040] Obstacle avoidance property is as follows: ,

[0041] If the path intersects with an obstacle, Taking infinity in the formula, that is... Because collisions with obstacles are strictly prohibited, a larger value should be given in practice depending on the situation.

[0042] The smaller the fitness value calculated in the end, the better the path.

[0043] (3) Calculate the adaptive nonlinear decay dynamic energy factor.

[0044] This implementation improves the energy factor in the artificial lemming algorithm, changing the original energy factor " Replace with an adaptive nonlinear decay dynamic energy factor:

[0045] in, To control for the impact of diversity, This is a nonlinear factor used to control time decay. This represents the maximum number of iterations. To normalize the initial diversity;

[0046] in, To prevent division by zero for the minimum value.

[0047] X(t) is a measure of population diversity at the t-th iteration:

[0048] Where N is the population size; Dim is the current dimension, where Dim represents the solution space of the problem. This invention relates to path planning in three-dimensional space, so Dim is 3, and each solution is (x, y, z). If it is a two-dimensional problem, Dim is 2, and each solution is (x, y). This represents the average value of all individuals in the k-th dimension at iteration number t; This represents the value of the k-th dimension of the i-th individual when the number of iterations is t.

[0049] X(0) represents the initial diversity of the population: , express The initial value; The population center is represented by a vector of average values ​​across all dimensions for all individuals in the current population.

[0050] wherein, represents the i-th individual of the t-th iteration; (4) according to the value of the dynamic factor E(t) to determine the next operation: a, if the energy factor E(t) is greater than or equal to 0.5, global exploration is carried out, Brownian motion simulation of irregular, random migration path is used by introducing adaptive Brownian step, and the difference between the current optimal solution and the random individual is updated; Global exploration is carried out, and the process of simulating irregular, random migration path (long-distance migration) by introducing adaptive Brownian step includes: Brownian motion is used to model, simulate irregular, random migration path, and update the difference between the current optimal solution and the random individual.

[0051] In the present application, adaptive Brownian step is introduced, so that the variance of Brownian step gradually decreases with iteration, large step exploration in early stage and small step refinement in later stage:

[0052] wherein, is the current individual path, which is actually represented by ; is the current optimal path; is a randomly selected individual in the population; is a direction control vector; is a parameter for changing the search direction; is a normal variable of Brownian motion, wherein , is the maximum step (the maximum distance of single jump in the search process of the algorithm); is the minimum step (the minimum distance of single jump in the search process of the algorithm).

[0053] The present application is based on the original is improved, and the in the original formula is replaced by adaptive Brownian step, that is, in the new formula. In this way, large step exploration in early stage and small step refinement in later stage can be realized, which is more suitable for actual path planning.

[0054] b, if 0.2 < E(t) < 0.5, a mixed strategy is adopted, and hole digging is carried out, that is, when the energy factor is between 0.2 and 0.5, search is carried out around the population center, balance development and exploration, simulate the behavior of traveling mice expanding randomly around a certain position, and construct a random coefficient to adjust the search direction based on a sine function.

[0055] This invention introduces a population center into the burrowing mechanism, and conducts regional exploration around the population center with a certain directionality, balancing the global and local directions.

[0056]

[0057] in, : Changes over time, creating volatility; rand is a random number between [0,1), F is the same as above. For random individuals; .

[0058] Since this is a hybrid strategy, this invention believes that the search should not continue to revolve around the optimal solution, but should achieve balance. Therefore, the original formula is modified accordingly. In Replace with .

[0059] c. If E(t)≤0.2, local exploitation is performed, foraging / evading pursuers. That is, when the energy factor is less than or equal to 0.2, the algorithm performs local exploitation, randomly adopting a spiral search strategy or a controlled Lévy flight strategy.

[0060] Foraging: Fine-tuning is performed by combining the current individual's positional relationship with the optimal solution using a spiral mechanism.

[0061] ,

[0062] in, This represents the Euclidean distance between the optimal individual and the current individual.

[0063] Escape from pursuers: Employ the Lévy flight mechanism to simulate escape behavior and develop a "long jump" strategy to avoid getting trapped in local optima.

[0064]

[0065] in, The square root decay method is used, which gradually decreases with each iteration to prevent the algorithm from jumping out of the optimal solution in the later stages:

[0066] in, The initial jump radius is determined based on the size of the actual problem space. t is the current iteration number. This represents the maximum number of iterations.

[0067] The original algorithm formula is After research, it is found that the Lévy flight of the original algorithm may still jump out of the neighborhood of the high-quality solution in the later stage, affecting convergence, therefore, in the application, the controlled Lévy flight is introduced to limit the jumping radius, so as to avoid jumping out of the excellent solution in the later stage.

[0068] (5) Boundary and obstacle checking is performed.

[0069] (6) Fitness of the individual (path) obtained in step (4) is evaluated, if the fitness of this iteration is better than the optimal fitness of the individual, the optimal individual is updated.

[0070] (7) If the fitness is better than the global optimal fitness, the global optimal is updated.

[0071] (8) Whether the maximum iteration number is reached is judged, if not, the next iteration is entered, if yes, the global optimal solution is output.

[0072] (9) Natural cubic spline interpolation is adopted to perform smoothing operation on the path, and the final output path is obtained.

[0073] The AUV motion path obtained through step (8) is a folded path connected by multiple straight line segments, such a path lacks smoothness, which is not feasible for the AUV. In order to make the controller keep stable and reduce energy consumption, and avoid exceeding the physical constraints of the AUV, the natural cubic spline interpolation is used to perform smoothing processing on the planned final path, that is, the path smoothing of the natural cubic spline interpolation, the embodiment does not improve the formula, only introduces the natural cubic spline interpolation to the application (MALA) at the application level. The specific process includes: The parameterized nodes obtained according to the path nodes of step (8) ; I represents the path point number, I=0,2,3,..,j.

[0074] Uniform parameterization is adopted:

[0075] Wherein represents the parameterized node value; Three spline interpolations are performed on each coordinate component, and the smoothed path points obtained by upsampling are obtained; the three spline interpolations in each interval are expressed by piecewise cubic polynomials:

[0076] Wherein, , the second derivative of the inner point satisfies the three-diagonal linear equation

[0077] u represents any interpolation parameter, that is, a continuous parameter variable at any point on the curve.

[0078] Splines are constructed for the x, y, and z components respectively, and their values ​​are evaluated on the parameter sampling set to obtain a smooth three-dimensional trajectory. . Example

[0079] To verify the effectiveness of the algorithm, this invention simulates the actual path planning of the algorithm in a complex 3D environment and compares it with three other algorithms. The comparison algorithms are Astar3D (3D 26-neighborhood A* algorithm), Goal-Biased RRT (goal bias RRT algorithm), and 3D-EACO-AH (3D adaptive heuristic ant colony algorithm).

[0080] Set the starting point as [10 50 1], the ending point as [350 240 100], and the straight-line distance between the starting point and the ending point as 401.8719m.

[0081] Comparison of effects is shown in Table 1 and Figures 2-4 As shown.

[0082] Table 1

[0083] Combination Figures 2-4 As can be seen, in complex 3D environments, the ALA algorithm can plan a safe path that satisfies the physical motion of the AUV and has a distance close to the shortest path in a relatively short time. The smoothed path is shown in the following diagram. Figures 8-9 As shown.

[0084] Change the start and end points to: Start point: [150 100 1], End point: [350 400 100], The straight-line distance from the start point to the end point is 373.8997.

[0085] Comparison of effects is shown in Table 2 and Figures 5-7 As shown in the image. The effect of path smoothing is as follows. Figures 10-11 As shown.

[0086] Table 2

[0087] Path planning in a realistic seabed environment: To verify the algorithm's effectiveness in actual marine environments, this invention utilizes simulation examples of real terrain. The terrain data comes from the Global Seabed Topography (GEBCO) map, geographically ranging from 38°S to 28°S and 154°E to 164°E. A three-dimensional spatial model is abstracted from the real environment to construct the underwater environment map. Figure ThreeThe V-grid model is used, and the starting point and the ending point are randomly selected to verify the feasibility of the algorithm.

[0088] In the experiment, the initial population parameter of the algorithm is set to 50, the maximum iteration number is 200, and two different starting points and ending points are randomly set: (1) The starting point is [50, -300, -1000], and the ending point is [-368, 432, -4000], and the effect is as shown in the figure. Figures 12-14

[0089] (2) The starting point is [-349, -410, -1000], and the ending point is [-203, 490, -4000], and the effect is as shown in the figure. Figures 15-17

[0090] The present application can also have other various embodiments, and those skilled in the art can make various corresponding changes and modifications according to the present application without departing from the spirit and essence of the present application, but these corresponding changes and modifications should belong to the protection scope of the claims attached to the present application.​​

Claims

1. An AUV path planning method based on a multi-strategy synergistic enhanced artificial mole algorithm, characterized in that: The method comprises the following steps: discretize the path into j intermediate control points , and the start point S and the end point G, to get the path; represent each candidate path individual as , and use the artificial ground squirrel algorithm to optimize the path to get the optimal path; The path optimization process using the artificial lemming algorithm comprises: In the initialization stage, the directional initialization strategy is adopted for initialization: taking the connection line of the starting point and the ending point as the center line, adding random disturbance to the interpolation points on the center line to generate path points, and then generating the initial population; selecting the smaller one as the superior path, calculating the self-adaptive nonlinear decay dynamic energy factor , is a weight factor for controlling the influence of diversity, is a nonlinear factor for controlling time decay; is the maximum number of iterations; is the normalized initial diversity; according to the value of the dynamic factor E(t), global exploration is performed, or hole digging is performed, or foraging / avoiding predators is performed; Then, boundary and obstacle checking is performed; the fitness of the current individual is evaluated, and if the fitness of the current iteration is better than the optimal fitness of the current individual, the optimal individual is updated; if the fitness is better than the global optimal fitness, the global optimal solution is updated; then, it is determined whether the maximum number of iterations is reached, if not, the next iteration is entered; if yes, the global optimal solution is output; and the final path is obtained based on the global optimal solution.

2. The AUV path planning method based on the multi-strategy synergistically enhanced artificial traveling salesman algorithm according to claim 1, characterized in that: Generate path points by adding random perturbation to the centerline interpolation points: , represents the jth path point of the ith individual; represents the jth path point of the base individual; perturbation represents the random perturbation value.

3. The AUV path planning method based on the multi-strategy synergistically enhanced artificial traveling salesman algorithm according to claim 2, characterized in that: Random perturbation value U is a matrix of size 1 3, where each element is a uniformly distributed random number in [0, 1]; N is the population size, i.e., the number of individuals in the population; UB is the upper bound of the three-dimensional coordinate, and LB is the lower bound of the three-dimensional coordinate.

4. The AUV path planning method based on the multi-strategy synergistically enhanced artificial traveling salesman algorithm according to claim 1, characterized in that: The path fitness function in the process of selecting the path with smaller fitness as the superior path is ， wherein, is the path length, is the turning angle, is the obstacle avoidance, is the weight of each index, which is set to ; represents the maximum path length, represents the maximum turning cost, represents the maximum obstacle avoidance risk.

5. The AUV path planning method based on the multi-strategy synergistically enhanced artificial traveling salesman algorithm according to claim 4, characterized in that: Obstacle avoidance As follows: , , i.e. infinity, a large value according to the practical situation.

6. The AUV path planning method based on the multi-strategy synergistically enhanced artificial traveling salesman algorithm according to any one of claims 1 to 5, characterized in that: Normalized initial diversity where, to prevent very small values except 0; N is the population size; Dim is the dimension of three-dimensional space; represents the average value of all individuals in the kth dimension at iteration t; represents the value of the kth dimension of the ith individual at iteration t; is the population center, which is the average value vector of all individuals in the current population in each dimension.

7. The AUV path planning method based on the multi-strategy synergistically enhanced artificial traveling salesman algorithm according to claim 6, characterized in that: The process of determining whether to perform global exploration, hole digging, or foraging / evading a pursuer based on the value of the dynamic factor E(t) comprises: a. If the energy factor E(t) is greater than or equal to a first threshold value, global exploration is performed, Brown motion simulation of irregular and random migration paths is used by introducing an adaptive Brown step length, and updating is performed in combination with the difference between the current optimal solution and a random individual; Global exploration is performed using a Brownian motion simulation that introduces irregular, random migration paths into the process with an adaptive Brownian step size where is the current individual path; is the current best path; is a randomly selected individual from the population; is the direction control vector; is a parameter used to change the search direction; is the normal variable of the Brownian motion, where , is the maximum step size; is the minimum step size; b. If the first threshold < E(t) < second threshold, a hybrid strategy is adopted, and a hole is dug, and a population center is introduced in the hole mechanism, and regional exploration is carried out around the population center, that is wherein, : manufacturing fluctuation changes over time; rand is a random number between [0, 1); F is the same as above; is a random individual; ; c. If E(t) is less than or equal to a second threshold value, foraging / evading a pursuer is performed.

8. The AUV path planning method based on the multi-strategy synergistically enhanced artificial traveling salesman algorithm according to claim 7, characterized in that: The process of foraging / evading a pursuer is as follows: Foraging: fine-tuning is performed by using a spiral mechanism in combination with the positional relationship between the current individual and the optimal solution; ， ， wherein, is the Euclidean distance between the best individual and the current individual; Evading a pursuer: escape behavior is simulated by using a Lévy flight mechanism to form a "long jump" strategy to avoid being trapped in a local optimum; ， wherein It gradually decreases with iteration using square root damping.

9. The AUV path planning method based on the multi-strategy synergistically enhanced artificial traveling salesman algorithm according to claim 8, characterized in that: The In, is the initial hop radius; t is the current iteration number; is the maximum iteration number.

10. The AUV path planning method based on the multi-strategy synergistically enhanced artificial traveling salesman algorithm according to claim 9, characterized in that: In the process of obtaining the final path based on the global optimal solution, a natural cubic spline interpolation method is used to smooth the path corresponding to the global optimal solution to obtain the final path.

Images