Multi-AUV obstacle avoidance path planning method based on chaotic mapping tiger-whale predation algorithm

Abstract

The invention discloses a multi-AUV obstacle avoidance path planning method based on a chaotic mapping tiger-whale predation algorithm, and the method comprises the following steps: (1) carrying out the three-dimensional modeling of an underwater environment, and building a virtual artificial potential field; (2) generating an initial population based on mixed chaotic mapping, and adaptively adjusting a mixing coefficient according to the barrier density; (3) iteratively optimizing the path by using a tiger whale predation algorithm, and dynamically avoiding obstacles in combination with obstacle avoidance constraint conditions; and (4) judging whether the current number of iterations reaches the maximum number of iterations, if a termination condition is met, stopping iteration, and outputting the optimized obstacle avoidance path of the AUV, otherwise, continuing iteration until the termination condition is met. According to the method, by means of randomness and ergodicity of chaotic mapping, the AUV can explore the surrounding environment more comprehensively without omission in the sensing detection link, and meanwhile, by means of the self-adaption characteristic, the obstacle avoidance path is flexibly adjusted in the avoidance action process.

Description

Technical Field

[0001] The present invention belongs to the technical field of underwater robot path planning, and in particular relates to a multi-AUV obstacle avoidance path planning method based on a chaotic mapping killer whale predation algorithm. Background Art

[0002] In recent years, with the increasing frequency of ocean exploration and development, autonomous underwater vehicles (AUVs) have become a research hotspot worldwide due to their high autonomy, flexibility, adaptability, and cost-effectiveness. Currently, single AUVs can perform a variety of underwater missions, such as resource exploration, reconnaissance, and counter-reconnaissance. However, when faced with complex operational missions, the limitations of single AUVs become apparent. For example, when tracking and locating unidentified, high-speed moving objects on the seabed, a single AUV can only rely on its limited tracking algorithms and power capabilities to attempt to lock onto the target. If the target's maneuvering strategy is complex and changeable, a single AUV, limited by its perception range and decision-making flexibility, often struggles to maintain and effectively track the target, resulting in a high probability of failure in the tracking and positioning mission. To address these issues, research is needed on AUV hardware performance or capture strategies. To overcome the shortcomings of a single AUV without incurring additional costs, the use of multiple AUV systems is currently the most ideal option.

[0003] In recent years, research on obstacle avoidance path planning for autonomous vehicles on land and in the air has been extensive, given factors such as relatively good communication conditions on land and in the air and the ease of control of autonomous vehicles. However, research on obstacle avoidance path planning for multiple underwater AUVs has lagged behind. Therefore, research on obstacle avoidance path planning for multiple underwater AUVs has profound strategic significance for enhancing China's future maritime capabilities.

[0004] To rationally assign path planning tasks to these AUVs, traditional collaborative control methods have traditionally been employed. Traditional collaborative control methods for multi-robot obstacle avoidance path planning primarily include precise optimization algorithms and heuristic algorithms. Precise optimization algorithms, such as branch-and-bound (BAB) and dynamic programming (DP), can achieve relatively accurate path planning results in simple scenarios. However, as the complexity of multi-AUV obstacle avoidance path planning increases, such as with the addition of constraints, the difficulty of solving the problem rapidly increases. Traditional heuristic algorithms, such as the particle swarm optimization (PSO) algorithm, which draws inspiration from the foraging behavior of bird flocks, the genetic algorithm (GA) inspired by Darwin's theory of evolution, and the ant colony optimization (ACO) algorithm, which was developed based on the phenomenon that ants leave pheromones behind them when foraging for food, which helps subsequent foraging ants navigate. Although these algorithms can reduce computational complexity to a certain extent, they still perform poorly in complex marine environments.

[0005] In summary, the methods described above are generally accepted and superior algorithms for solving the multi-AUV obstacle avoidance path planning problem. However, traditional algorithms are prone to entering local optimal solutions, suffer from poor convergence, and suffer from unstable search results. The proposed multi-AUV obstacle avoidance path planning method based on the chaotic mapping killer whale predation algorithm simulates killer whale predation behavior, exploits the randomness and ergodicity of chaotic mapping, and adaptively adjusts the AUV's path to avoid obstacles. This method largely overcomes the obstacle avoidance and path planning limitations of traditional algorithms, which are prone to entering local optimal solutions, thereby enabling better decision-making. Summary of the Invention

[0006] The purpose of the present invention is to solve the defect that the traditional algorithm used in the existing multi-AUV obstacle avoidance path planning is prone to fall into local optimality, and propose a multi-AUV obstacle avoidance path planning method based on the chaotic mapping killer whale predation algorithm.

[0007] For the problem of multi-AUV obstacle avoidance path planning, in view of the fact that traditional algorithms need to manually formulate constraints and have a complex solution process when dealing with complex environments, in-depth optimization research is carried out based on the killer whale predation algorithm based on chaos mapping. First, the underwater environment is modeled by 3D rasterization and virtual potential field fusion. This environment highly restores the various shapes and irregular distribution of obstacles that exist in the real underwater environment. In response to this complex environment, an improved killer whale predation algorithm that integrates adaptive strategy and chaotic mapping mechanism is innovatively proposed. This algorithm uses the randomness and ergodicity of chaotic mapping to enable AUVs to explore the surrounding environment more comprehensively and without omissions during perception and detection. At the same time, it uses adaptive characteristics to flexibly adjust the obstacle avoidance path during avoidance actions.

[0008] In order to achieve the above object, the technical solution of the present invention is as follows:

[0009] A multi-AUV obstacle avoidance path planning method based on a chaotic mapping killer whale predation algorithm includes the following steps:

[0010] (1) Conduct three-dimensional modeling of the underwater environment and establish a virtual artificial potential field:

[0011] The grid method is used to segment the underwater environment to form a three-dimensional grid map. Each grid is marked according to whether there are obstacles (static reefs, dynamic ocean current areas, etc.), and which grids are obstacles and which are movable spaces. A three-dimensional array is used to store map data. Set the starting point S(x s ,y s ,z s ) and the end point T(x t ,y t ,z t ).

[0012] Construct a virtual artificial potential field, use the gravitational potential function to simulate the gravitational field between the agent and the target point, and use the repulsive potential function to simulate the repulsive field between the agent and the obstacle.

[0013] The gravitational field is designed as follows:

[0014]

[0015] Where k a represents the gravitational field gain coefficient, ρ(q,q g ) represents the position of the AUV in the world coordinate system q = (x, y) and the target position q g =(x g ,y g ) between the distance vector, the size is the Euclidean distance |qq g |, the direction points to the target location. The magnitude of the gravitational force is the negative gradient of the gravitational field, that is:

[0016]

[0017] For example, to solve the problem that the AUV oscillates at the target point or cannot reach the target point, the repulsive field is designed as:

[0018]

[0019] Where k r represents the repulsive field gain coefficient, h is an adjustable positive coefficient, ρ(q,q ob ) is measured by the underwater acoustic sensor, indicating the distance vector sum between the AUV and the obstacle, with the direction pointing to the AUV, l is the maximum measurement distance of the distance sensor, ρ g is the distance between the AUV and the target point. The repulsive force is the negative gradient of the repulsive field, and its expression is:

[0020] F r (q)=F rl (q)+F r2 (q)

[0021]

[0022] Where, F rl The direction from the obstacle to the AUV, F r2 The direction of the AUV is pointed towards the target point.

[0023] In order to cope with the time-varying ocean currents underwater, the ocean current force is also added to the construction of the artificial potential field:

[0024]

[0025] Where: ρ(q o ,q g) represents the distance from the origin to the end point, V c represents the ocean current vector; F c (q) As the AUV approaches the target point, the impact gradually decreases, so that when the AUV is closer to the target point, it can overcome the influence of the ocean current and reach the target point faster. Therefore, the resultant force on the AUV is:

[0026] F=F a (q)+F c (q)+T(F r (q))

[0027] =F a (q)+F c (q)+T(F r1 (q)+F r2 (q))′

[0028] (2) Generate the initial population based on the mixed chaotic map and adaptively adjust the mixing coefficient according to the obstacle density:

[0029] The chaotic map of linear superposition of Tent map and Logistic map is used to generate chaotic sequence. The expression of mixed chaotic map is:

[0030] C hybrid (t) = β·C Tent (t)+(1-β)·C Logistic (t)

[0031] Among them, β is the mixing coefficient, according to the environmental obstacle density ρ obs Adaptive Adjustment:

[0032]

[0033] Where, ρ threshold is the obstacle density threshold. The mixed chaotic sequence is mapped to three-dimensional space to generate an initialized population, where each individual in the population represents a potential path point for an AUV.

[0034] (3) Use the killer whale predation algorithm to iteratively optimize the path, combined with obstacle avoidance constraints, to dynamically avoid obstacles:

[0035] The specific steps of iteratively optimizing the path using the killer whale predation algorithm in step (3) are as follows:

[0036] (3-1) Evaluate the fitness function value of each killer whale and select the best killer whale individual and its position. The fitness function consists of the objective function and the constraint function, which can be expressed as:

[0037] F=ω1R o +ω2R c

[0038] Among them, ω1 and ω1 are weight coefficients, R o is the objective function, which is the potential field reward function proposed based on the artificial potential field and can be expressed as:

[0039]

[0040] Where, F′=F / ||F||, F′ x , F′ y is the component of F′ in X and Y, ψ is the yaw of AUV, and r∈R. The constraint function is the obstacle avoidance reward function, which can be expressed as:

[0041] R c =r o +r b +r e +r f

[0042] Where r o is the safety obstacle avoidance reward, which can be expressed as:

[0043]

[0044] in, and is a constant, E r Indicates whether the AUV has entered the obstacle range. The closer the AUV is to the obstacle, the less reward it gets. and The closer the AUV's heading direction (including pitch and yaw angles) is to the obstacle, the smaller the reward.

[0045] r b The reward for reaching the target area can be expressed as:

[0046]

[0047] in, is r b In addition, in order to ensure that the action of AUV approaching the target area is always rewarded rather than punished, the weight value must be satisfied when setting the weight value.

[0048] r e The shortest path reward can be expressed as:

[0049]

[0050] in, is r e The setting weight, τ u-max ,τ q-max ,τr-max Represents the maximum value of the three control signals and sets the reward value of each time step to a negative value, which means that the AUV can reach the specified target area with a shorter time step and thus obtain a larger reward.

[0051] r f Is the reward for completing the obstacle avoidance task, which can be expressed as:

[0052]

[0053] Where A represents the terminal state, A=[0,0] T Indicates that the AUV navigates normally in the mission area and does not collide. A=[0,1] T Indicates that the number of iterations reaches the set value, A=[0,1] T Indicates that the AUV runs out of the mission area or collides with an obstacle, A=[1,1] T Arrive at the designated target area without collision.

[0054] (3-2) Killer whales choose to chase or surround the school of fish based on the selection factor p1. During this period, they use sonar to locate the prey and adjust their position.

[0055] The selection factor p1 is a random constant between [0,1], and the numbers between [0,1] are also randomly generated. When the value of the number is greater than p1, the school of fish will be driven away, otherwise the school of fish will be surrounded.

[0056] During the fish-herding phase, the speed and position of the killer whale are expressed by the following formula:

[0057]

[0058] c=1-b

[0059]

[0060] Where M represents the average position of the killer whale group, is the position of the killer whale with the smallest fitness value in the group at the tth iteration, is the current position of the i-th killer whale at the t-th iteration, N is the total number of killer whales, M t is the average value of all killer whale positions at the tth iteration (group center position), is the position of the i-th killer whale after being updated by the first driving strategy, is the position of the i-th killer whale after being updated by the second driving strategy, t represents the number of cycles, a, b, d are random values between [0, 1], e is a random value between [0, 2], and q is a number between [0, 1], which represents the probability of selecting a certain driving strategy.

[0061] During the fish encirclement phase, killer whales communicate with each other to determine their next location. During this period, the position adjustment of killer whales is expressed by the following formula:

[0062]

[0063] Where T is the maximum number of iterations of the algorithm, u is the dynamic adjustment coefficient of the encirclement phase that decreases nonlinearly with the number of iterations. represents the position of the i-th killer whale individual after choosing the third pursuit method at the t-th iteration, are the positions of three different killer whale individuals randomly selected in the kth dimension at the tth iteration (j1≠j2≠j3≠i);

[0064] During these two phases, the killer whales sense the position of the school of fish and adjust their own position accordingly. Their position is adjusted according to the following formula:

[0065]

[0066] Where f(x) is the fitness function, x corresponds to the position of the killer whale individual, and the smaller the value of the fitness function, the closer the corresponding killer whale position is to the target. is the position of the i-th killer whale individual at the t-th iteration of the pursuit phase, is the current position of the i-th killer whale at the t-th iteration. Represents The corresponding fitness function value is, Represents The corresponding fitness function value.

[0067] (3-3) During the attack phase, killer whales prey by attacking schools of fish. During this phase, they update their positions using sonar, and the positions of some killer whales are replaced by the lower limit lb of the search space. During this period, the changes in the killer whales' positions and speeds can be expressed as follows:

[0068]

[0069] Where, represents the four best-positioned killer whales, j1, j2, and j3 represent three different killer whales randomly selected during the pursuit phase, g1 is a random number between [0, 2], and g2 is a random number between [-2.5, 2.5]. The positions of some killer whales in poor positions are replaced with the lower boundary of the search space, lb.

[0070] (3-4) Each AUV shares its own path information based on the underwater acoustic communication network and coordinates the path according to the relative position relationship with other AUVs.

[0071] Each AUV shares its path information in real time through the underwater acoustic communication network, including its current position, planned path, and fitness value. After receiving information from other AUVs, each AUV analyzes the relative positions of multiple AUVs and makes coordinated position adjustments to move toward the target location and avoid collisions.

[0072] (4) Determine whether the current number of iterations has reached the maximum number of iterations. If the termination condition is met, stop the iteration and output the optimized obstacle avoidance path of the AUV. Otherwise, continue the iteration until the termination condition is met.

[0073] Compared with the prior art, the present invention has the following beneficial effects:

[0074] 1. The chaotic mapping killer whale predation algorithm adopted by the method of the present invention has a faster convergence speed than the traditional killer whale predation algorithm, which speeds up the program running speed.

[0075] 2. Adaptive hybrid chaotic mapping is used to generate the initial population. By combining the Logistic Chaotic Map and the Tent Chaotic Map and dynamically adjusting the mixing coefficient based on obstacle density, the generated initial path solutions are more evenly and widely distributed across the search space. In areas with dense obstacles, the Logistic Chaotic Map weight is increased to generate initial paths that better fit the complex environment; in sparse areas, the Tent Chaotic Map is enhanced to explore the space more extensively. This greatly increases the diversity of the initial population, providing a richer and higher-quality initial solution for subsequent optimization, giving the algorithm a greater chance of escaping local optima and finding the global optimal solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 Schematic diagram of artificial potential field;

[0077] Figure 2 Schematic diagram of underwater environment rasterization;

[0078] Figure 3 Flowchart of generating initial population by hybrid chaotic mapping;

[0079] Figure 4 Flowchart of iterative optimization path of killer whale predation algorithm;

[0080] Figure 5 Schematic diagram of the killer whale's predation algorithm's obstacle avoidance. DETAILED DESCRIPTION

[0081] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention.

[0082] A multi-AUV obstacle avoidance path planning method based on a chaotic mapping killer whale predation algorithm includes the following steps:

[0083] (1) Conduct three-dimensional modeling of the underwater environment and establish a virtual artificial potential field;

[0084] (2) Generate the initial population based on the mixed chaotic map, adaptively adjust the mixing coefficient according to the obstacle density, and perform algorithm initialization settings;

[0085] (3) Use the killer whale predation algorithm to iteratively optimize the path, combined with obstacle avoidance constraints, and dynamically avoid obstacles;

[0086] (4) Determine whether the current number of iterations has reached the maximum number of iterations. After the termination condition is met, output the optimized obstacle avoidance path of the AUV.

[0087] like Figure 1 As shown in the above step (1), the grid method is used to segment the underwater environment to form a three-dimensional spatial grid map. Each grid is marked according to whether there are obstacles (static reefs, dynamic ocean current areas, etc.). The grid with obstacles is marked as 1 (Occupied) and the movable space is marked as 0 (Free). Use a three-dimensional array to store map data. Set the starting point S(x s ,y s ,z s ) and the target point T(x t ,y t ,z t ).

[0088] Construct a virtual artificial potential field, use the gravitational potential function to simulate the gravitational field between the agent and the target point, and use the repulsive potential function to simulate the repulsive field between the agent and the obstacle.

[0089] The gravitational field is designed as follows:

[0090]

[0091] Where k a represents the gravitational field gain coefficient, ρ(q,q g ) represents the position of the AUV in the world coordinate system q = (x, y) and the target position q g =(x g ,y g ) between the distance vector, the size is the Euclidean distance |qq g |, the direction points to the target location. The magnitude of the gravitational force is the negative gradient of the gravitational field, that is:

[0092]

[0093] For example, to solve the problem that the AUV oscillates at the target point or cannot reach the target point, the repulsive field is designed as:

[0094]

[0095] Where k r represents the repulsive field gain coefficient, h is an adjustable positive coefficient, ρ(q,q ob ) is measured by the underwater acoustic sensor, indicating the distance vector sum between the AUV and the obstacle, with the direction pointing to the AUV, l is the maximum measurement distance of the distance sensor, ρ g is the distance between the AUV and the target point. The repulsive force is the negative gradient of the repulsive field, and its expression is:

[0096] F r (q)=F rl (q)+F r2 (q)

[0097]

[0098] Where, F rl The direction from the obstacle to the AUV, F r2 The direction of the AUV is pointed towards the target point.

[0099] In order to cope with the time-varying ocean currents underwater, the ocean current force is also added to the construction of the artificial potential field:

[0100]

[0101] Where: ρ(q o ,q g ) represents the distance from the origin to the end point, F c represents the ocean current vector; F c (q) As the AUV approaches the target point, the influence gradually decreases, so that the AUV can overcome the influence of the ocean current and reach the target point faster when it is closer to the target point. Therefore, the resultant force on the AUV is as follows: Figure 2 As shown, its value is:

[0102] F=F a (q)+F c (q)+T(F r (q))

[0103] =F a (q)+F c (q)+T(F r1 (q)+F r2 (q))′

[0104] like Figure 3As shown in the above step (2), the chaotic map of linear superposition of Tent map and Logistic map is used to generate chaotic sequence. The expression of mixed chaotic map is:

[0105] C hybrid (t) = β·C Tent (t)+(1-β)·C Logistic (t)

[0106] Among them, β is the mixing coefficient, according to the environmental obstacle density ρ obs Adaptive Adjustment:

[0107]

[0108] Where, ρ threshold is the obstacle density threshold, ρ threshold = 0.3. The mixed chaotic sequence is mapped to three-dimensional space and converted into the AUV's initial path solution, expressed as node coordinates, where each coordinate corresponds to a position on the grid map. After mapping, the starting point S and target point T are forcibly corrected to ensure the AUV moves toward the target.

[0109] The algorithm parameters are set as follows: population size N = 50, maximum number of iterations T = 200, and initial mixing coefficient β = 0.5.

[0110] like Figure 4 As shown, the specific steps of iteratively optimizing the path using the killer whale predation algorithm in step (3) are:

[0111] (3-1) The fitness function value of each killer whale is evaluated in combination with the obstacle avoidance constraint, and the best killer whale individual and its location are selected. That is, the fitness function value of each path of each AUV is calculated, and the AUV with the smallest fitness function value and its corresponding path are selected.

[0112] The fitness function consists of the objective function and the constraint function, which can be expressed as:

[0113] F=ω1R o +ω2R c

[0114] Among them, ω1 and ω2 are weight coefficients, R o is the objective function, which is the potential field reward function proposed based on the artificial potential field and can be expressed as:

[0115]

[0116] Where, F′=F / ||F||, F′ x , F′ yis the component of F′ in X and Y, ψ is the yaw of AUV, and r∈R. The constraint function is the obstacle avoidance reward function, which adds obstacle avoidance constraints to the fitness function and enables AUV to achieve dynamic obstacle avoidance, such as Figure 5 As shown. The constraint function can be expressed as:

[0117] R c =r o +r b +r e +r f

[0118] Where r o is the safety obstacle avoidance reward, which can be expressed as:

[0119]

[0120] in, and is a constant, E r Indicates whether the AUV has entered the obstacle range. The closer the AUV is to the obstacle, the less reward it gets. and The closer the AUV's heading direction (including pitch and yaw angles) is to the obstacle, the smaller the reward.

[0121] r b The reward for reaching the target area can be expressed as:

[0122]

[0123] in, is r b In addition, in order to ensure that the action of AUV approaching the target area is always rewarded rather than punished, the weight value must be satisfied when setting the weight value. r e The shortest path reward can be expressed as:

[0124]

[0125] in, is r e The setting weight, τ u-max ,τ q-max ,τ r-max Represents the maximum value of the three control signals and sets the reward value of each time step to a negative value, which means that the AUV can reach the specified target area with a shorter time step and thus obtain a larger reward.

[0126] r f Is the reward for completing the obstacle avoidance task, which can be expressed as:

[0127]

[0128] Where A represents the terminal state, A=[0,0] T Indicates that the AUV navigates normally in the mission area and does not collide. A=[0,1] T Indicates that the number of iterations reaches the set value, A=[0,1] T Indicates that the AUV runs out of the mission area or collides with an obstacle, A=[1,1] T Arrive at the designated target area without collision.

[0129] (3-2) Killer whales choose to chase or surround the school of fish based on a selection factor, p1. During this process, they use sonar to locate the prey and adjust their own position. Each AUV determines a different forward motion pattern based on a random selection factor and shares its path information through an underwater acoustic communication network.

[0130] The selection factor p1 is a random constant between [0, 1], and the numbers between [0, 1] are also randomly generated. When the value of the number is greater than p1, the school of fish will be driven away; otherwise, the school of fish will be surrounded. During the stage of driving away the school of fish, the speed and position of the killer whale can be expressed by the following formula:

[0131]

[0132] c=1-b

[0133]

[0134] Where M represents the average position of the killer whale group, is the position of the killer whale with the smallest fitness value in the group at the tth iteration, is the current position of the i-th killer whale at the t-th iteration, N is the total number of killer whales, M t is the average value of all killer whale positions at the tth iteration (group center position), is the position of the i-th killer whale after being updated by the first driving strategy, is the position of the i-th killer whale after being updated by the second driving strategy, t represents the number of cycles, a, b, d are random values between [0, 1], e is a random value between [0, 2], and q is a number between [0, 1], which represents the probability of selecting a certain driving strategy.

[0135] During the fish encirclement phase, killer whales communicate with each other to determine their next location. During this period, the position adjustment of killer whales is expressed by the following formula:

[0136]

[0137] Where T is the maximum number of iterations of the algorithm, u is the dynamic adjustment coefficient of the encirclement phase that decreases nonlinearly with the number of iterations. represents the position of the i-th killer whale individual after choosing the third pursuit method at the t-th iteration, are the positions of three different killer whale individuals randomly selected in the kth dimension at the tth iteration (j1≠j2≠j3≠i).

[0138] During these two phases, the killer whales sense the position of the school of fish and adjust their own position accordingly. Their position is adjusted according to the following formula:

[0139]

[0140] Where f(x) is the fitness function, x corresponds to the position of the killer whale individual, and the smaller the value of the fitness function, the closer the corresponding killer whale position is to the target. is the position of the i-th killer whale individual at the t-th iteration of the pursuit phase, is the current position of the i-th killer whale at the t-th iteration. Represents The corresponding fitness function value is, Represents The corresponding fitness function value.

[0141] (3-3) During the attack phase, killer whales prey by attacking schools of fish. During this phase, they update their positions using sonar, and the positions of some killer whales are replaced by the lower limit lb of the search space. This means that the AUV continues to update its path information through the underwater acoustic network. During this period, the AUV's speed and position changes can be expressed as follows:

[0142]

[0143]

[0144] Where, represents the four AUVs at the best position, j1, j2, and j3 represent three randomly selected AUVs, g1 is a random number between [0, 2], and g2 is a random number between [-2.5, 2.5]. The positions of some AUVs with poor positions are replaced with the lower boundary lb of the search space.

[0145] The pursuit and attack phases here are caused by the changes in the current optimal path points when multiple AUVs approach the target point.

[0146] (3-4) Each AUV shares its own path information based on the underwater acoustic communication network and coordinates the path according to the relative position relationship with other AUVs.

[0147] Each AUV shares its path information in real time through the underwater acoustic communication network, including its current position, planned path, and fitness value. After receiving information from other AUVs, each AUV analyzes the relative positions of multiple AUVs and makes coordinated position adjustments to move toward the target location and avoid collisions.

[0148] Step (4) above is the termination condition judgment, which determines whether the current number of iterations t reaches the maximum number of iterations T. If the termination condition is met, all optimal path points are connected and smoothed using a B-spline curve, and the optimized path planning result is output. Otherwise, the process goes to step (3-1).

[0149] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A multi-AUV obstacle avoidance path planning method based on chaos mapping killer whale predation algorithm, characterized in that: The steps include: (1) Conduct three-dimensional modeling of the underwater environment and establish a virtual artificial potential field; (2) Generate the initial population based on the mixed chaotic map and adaptively adjust the mixing coefficient according to the obstacle density; (3) Use the killer whale predation algorithm to iteratively optimize the path, combined with obstacle avoidance constraints, and dynamically avoid obstacles; (4) Determine whether the current number of iterations has reached the maximum number of iterations. If the termination condition is met, stop the iteration and output the optimized obstacle avoidance path of the AUV. Otherwise, continue the iteration until the termination condition is met.

2. The multi-AUV obstacle avoidance path planning method based on the chaotic mapping killer whale predation algorithm according to claim 1 is characterized in that: The specific steps of step (1) are as follows: The grid method is used to segment the underwater environment to form a three-dimensional grid map. Each grid is marked according to whether there are obstacles. It is determined which grids are obstacles and which spaces can be moved. The map data is stored in a three-dimensional array and the starting point S(x s ,y s ,z s ) and the end point T(x t ,y t ,z t ); Construct a virtual artificial potential field, using the gravitational potential function to simulate the gravitational field between the agent and the target point, and the repulsive potential function to simulate the repulsive field between the agent and the obstacle; The gravitational field is designed as follows: Where k a represents the gravitational field gain coefficient, ρ(q,q g ) represents the position of the AUV in the world coordinate system q = (x, y) and the target position q g =(x g ,y g ) between the distance vector, the size is the Euclidean distance |qq g |, the direction points to the target position; the magnitude of the gravitational force is the negative gradient of the gravitational field, that is: In order to solve the problem that the AUV oscillates at the target point or cannot reach the target point, the repulsive field is designed as follows: Where k r represents the repulsive field gain coefficient, h is an adjustable positive coefficient, ρ(q,q ob ) is measured by the underwater acoustic sensor, indicating the distance vector sum between the AUV and the obstacle, with the direction pointing to the AUV, l is the maximum measurement distance of the distance sensor, ρ g is the distance between the AUV and the target point; the repulsive force is the negative gradient of the repulsive field, and its expression is: F r (q)=F rl (q)+F r2 (q) Where, F rl The direction from the obstacle to the AUV, F r2 The direction of the AUV is pointed to the target point; In order to cope with the time-varying ocean currents underwater, the ocean current force is also added to the construction of the artificial potential field: Where: ρ(q o ,q g ) represents the distance from the origin to the end point, V c represents the ocean current vector; F c (q) As the AUV approaches the target point, the impact gradually decreases, so that when the AUV is closer to the target point, it can overcome the influence of the ocean current and reach the target point faster. The resultant force on the AUV is:

3. The multi-AUV obstacle avoidance path planning method based on the chaotic mapping killer whale predation algorithm according to claim 1 is characterized in that: The specific steps of step (2) are as follows: The chaotic map of linear superposition of Tent map and Logistic map is used to generate chaotic sequence. The expression of mixed chaotic map is: C hybrid (t)=β·C Tent (t)+(1-β)·C Logistic (t) Among them, β is the mixing coefficient, according to the environmental obstacle density ρ obs Adaptive Adjustment: Where, ρ threshold is the obstacle density threshold, and the mixed chaotic sequence is mapped to three-dimensional space to generate an initialized population. Each individual in the population represents a potential path point for an AUV.

4. The multi-AUV obstacle avoidance path planning method based on the chaotic mapping killer whale predation algorithm according to claim 1 is characterized in that: The specific steps of iteratively optimizing the path using the killer whale predation algorithm in step (3) are as follows: (3-1) Evaluate the fitness function value of each killer whale and select the best killer whale individual and its position. The fitness function consists of the objective function and the constraint function, which is expressed as: F=ω1R o +ω2R c Among them, ω1 and ω1 are weight coefficients, R o is the objective function, which is the potential field reward function proposed based on the artificial potential field and is expressed as: Where, F′=F / ||F||, F′ x , F′ y is the component of F′ on X and Y, ψ is the yaw of AUV, r∈R; the constraint function is the obstacle avoidance reward function, which is expressed as: R c ＝r o +r b +r e +r f Where r o is the safety obstacle avoidance reward, expressed as: in, and is a constant, E r Indicates whether the AUV has entered the obstacle range. The closer the AUV is to the obstacle, the less reward it gets. and The more the AUV moves towards the obstacle, the less reward it will receive. r b Reward for reaching the target area, expressed as: in, is r b To ensure that the AUV is always rewarded rather than punished for approaching the target area, the weight value must be satisfied when setting the weight value. r e The shortest path reward is expressed as: in, is r e The setting weight, τ u-max ,τ q-max ,τ r-max Represents the maximum value of the three control signals. Setting the reward value of each time step to a negative value means that the AUV can reach the specified target area in a shorter time step, thereby obtaining a larger reward; r f Is the reward for completing the obstacle avoidance task, expressed as: Where A represents the terminal state, A=[0,0] T Indicates that the AUV navigates normally in the mission area and does not collide. A=[0,1] T Indicates that the number of iterations reaches the set value, A=[0,1] T Indicates that the AUV runs out of the mission area or collides with an obstacle, A=[1,1] T Arrived at the designated target area without collision; (3-2) Killer whales choose to drive away or surround the school of fish according to the selection factor p1 to complete the chasing process. During this period, they locate the prey and adjust their own position through sonar. The selection factor p1 is a random constant between [0,1], and the numbers between [0,1] are also randomly generated. When the value of the number is greater than p1, the school of fish will be driven away, otherwise the school of fish will be surrounded. During the fish-herding phase, the speed and position of the killer whale are expressed by the following formula: c=1-b Where M represents the average position of the killer whale group, is the position of the killer whale with the smallest fitness value in the group at the tth iteration, is the current position of the i-th killer whale at the t-th iteration, N is the total number of killer whales, M t is the average of all killer whale positions at the tth iteration, is the position of the i-th killer whale after being updated by the first driving strategy, is the position of the i-th killer whale after the second driving strategy is updated, t represents the number of cycles, a, b, d are random values between [0, 1], e is a random value between [0, 2], and q is a number between [0, 1], representing the probability of selecting a certain driving strategy; During the fish encirclement phase, killer whales communicate with each other to determine their next location. During this period, the position adjustment of killer whales is expressed by the following formula: Where T is the maximum number of iterations of the algorithm, u is the dynamic adjustment coefficient of the encirclement phase that decreases nonlinearly with the number of iterations. represents the position of the i-th killer whale individual after choosing the third pursuit method at the t-th iteration, are the positions of three different killer whale individuals randomly selected in the kth dimension at the tth iteration (j1≠j2≠j3≠i); During these two phases, the killer whales sense the position of the school of fish and adjust their own position accordingly. Their position is adjusted according to the following formula: Where f(x) is the fitness function, x corresponds to the position of the killer whale individual, and the smaller the value of the fitness function, the closer the corresponding killer whale position is to the target; is the position of the i-th killer whale individual at the t-th iteration of the pursuit phase, is the current position of the i-th killer whale at the t-th iteration; Represents The corresponding fitness function value is, Represents The corresponding fitness function value; (3-3) During the attack phase, killer whales prey by attacking schools of fish. During this period, they update their own positions through sonar, and the positions of some killer whales are replaced by the lower limit lb of the search space. During this period, the changes in the position and speed of killer whales are expressed by the following formula: Where, represents the four killer whales in the best position, j1, j2, and j3 represent three different killer whales randomly selected during the pursuit phase, g1 is a random number between [0, 2], and g2 is a random number between [-2.5, 2.5]. The positions of some killer whales with poor positions are replaced with the lower boundary lb of the search space. (3-4) Each AUV shares its own path information based on the underwater acoustic communication network, including the current position, planned path and fitness, and coordinates the path according to the relative position relationship with other AUVs; after each AUV receives information from other AUVs, it analyzes the relative position relationship between multiple AUVs and performs coordinated position adjustment to move towards the target position and avoid collision.

Images