Mineral collection task scheduling method and system for multiple underwater robots
Through the combination of particle swarm algorithm and three-dimensional current field model, the underwater mineral collection task path is optimized, the problem of inefficiency in the existing technology is solved, and efficient mineral collection task allocation and execution in multi-underwater robot systems are realized.
Patent Information
- Application Number
- CN202510043634.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-10
- Publication Date
- 2025-05-30
AI Technical Summary
The existing underwater mineral collection task methods are inefficient and difficult to meet the needs of large-scale tasks, especially in multi-underwater robot systems. Distance optimization is insufficiently considered separately and it is difficult to efficiently plan in complex ocean current environments.
The particle swarm algorithm is used to combine the three-dimensional current field model to optimize the optimal path between different mineral collection points, and allocate mineral collection point sequences to multiple underwater robots through layered planning to improve the efficiency of the collection task.
In the complex ocean current environment, the efficiency of underwater mineral collection tasks is significantly improved, the reasonable and efficient task allocation of multiple robots is achieved, energy consumption is reduced, and the robustness and practicality of the system is improved.
Smart Images
Figure CN120066092A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of mobile robots, and particularly relates to a mineral collection task scheduling method and system for multiple underwater robots. Background Art
[0002] An autonomous underwater vehicle (AUV) is a task controller that integrates artificial intelligence and other advanced computing technologies. With the continuous maturity of its technology development, it has become an underwater autonomous operation system capable of completing specified tasks. However, for a single AUV, its functions and capacities are very limited and it is difficult to meet large-scale task requirements. Therefore, the concept of a multi-autonomous underwater vehicle system (MAUV) has emerged. The MAUV system is an important direction for the development of underwater robot technology. Through the cooperation and coordination among multiple robots, it can not only enhance the basic functions of each robot, but also expand more intelligent behaviors during the interaction process. This system plays an increasingly important role in tasks such as three-dimensional cooperative detection in underwater environments, underwater cooperative search, underwater enclosure, and underwater mineral collection. Its development helps to improve the intelligent level of underwater robots and accelerate the research and development process of marine equipment.
[0003] Underwater mineral collection is one of the common working scenarios of AUVs. Currently, when underwater robots execute mineral collection tasks, they often traverse the collection task points from near to far according to the distance. Although this method can traverse all collection task points in the working space in order, it only considers distance as an optimization index and has the problem of low efficiency. At the same time, this method often only involves the planning of a single AUV. When operating in a large underwater scene, the feasibility of this method is relatively low and it is difficult to meet task requirements. Summary of the Invention
[0004] The purpose of the present invention is to propose a mineral collection task scheduling method and system for multiple underwater robots to improve the efficiency of underwater mineral collection in view of the deficiencies of the current mineral collection task methods.
[0005] In the first aspect, the present invention provides a mineral collection task scheduling method for multiple underwater robots, which includes the following steps: Step 1: Construct a three-dimensional ocean current field model; Step 2: Use the particle swarm algorithm to obtain the optimal paths between different mineral collection points respectively. In each iteration process, introduce the constructed three-dimensional ocean current field model to update the position and fitness of each particle; Step 3: Assign a sequence of mineral collection points to the underwater robots; take the optimal collection order in the sequence of mineral collection points corresponding to the underwater robots as the target sequence, obtain the target sequences of all underwater robots respectively according to the optimal paths between different mineral collection points, and use the target sequences to complete the mineral collection task scheduling of the underwater robots.
[0006] Preferably, in the second step, the method for updating the position of each particle is as follows: Wherein, are the position components of the d-th particle after i-th iteration in the x, y, and z directions respectively; are the velocity components of the d-th particle after i-th iteration along the x, y, and z axes respectively; , , are the eddy current velocity components in the horizontal, vertical, and vertical axis directions respectively; ; D is the size of the particle swarm; is the current iteration number.
[0007] Preferably, in the second step, the method for updating the fitness of each particle is as follows: Wherein, is the fitness of the particle; and are the path loss and ocean current loss respectively; and are the path loss weight and ocean current loss weight respectively; are the x, y, and z axis coordinates of the r-th path point in the particle; R is the number of path points; is the friction coefficient of seawater; is the distance from the current path point to the next path point of the underwater robot; V is the speed of the underwater robot; D is the diameter of the underwater robot; is the acceleration due to gravity.
[0008] Preferably, in the second step, the iterative update method of the individual learning factor and the global learning factor in the particle swarm algorithm is as follows: Wherein, is the individual learning factor; is the global learning factor; and are the initial value and the final value of the individual learning factor respectively; and are the initial value and the final value of the global learning factor respectively; is the current iteration number; is the maximum iteration number.
[0009] Preferably, in the third step, the method for obtaining the target sequence of the underwater robot is as follows: taking different permutations of the mineral collection points in the collection point sequence as the initial population; continuously iterating the processes of selection, crossover, mutation, and simulated annealing, and regenerating the initial population before each iteration until the number of iterations reaches the maximum number of iterations; selecting the individual with the lowest total path cost among all iterations as the target sequence.
[0010] Preferably, the method for obtaining the total path cost is as follows: taking the fitness corresponding to the optimal path between different mineral collection points as its path cost, and adding the path costs between different mineral collection points in sequence according to the arrangement order of the mineral collection points in the collection point sequence to obtain the total path cost.
[0011] Preferably, the method of simulated degradation is as follows: Randomly select an individual from the population obtained after mutation, and use the two-element optimization method to transform the path of the individual; if the total path cost of the transformed individual is less than that of the original individual, replace the original individual with the transformed individual and add it to the population; otherwise, with a probability Replace the original individual with the transformed individual and add it to the population; where is the difference between the total path costs of the transformed individual and the original individual; is the current temperature; loop this process until the number of loops k is greater than the maximum number of loops K.
[0012] Preferably, in the first step, the method for constructing the three-dimensional ocean current field model is as follows: using the Navier-Stokes equation to construct a viscous Lamb vortex field, and superimposing multiple Lamb vortices to obtain the ocean current field model.
[0013] Preferably, in the third step, the mineral collection point sequences corresponding to different underwater robots do not intersect each other and their sum covers all mineral collection points.
[0014] In a second aspect, the present invention provides a mineral collection task scheduling system for multiple underwater robots, which is used to execute the above-mentioned mineral collection task scheduling method; the mineral collection task scheduling system includes an ocean current field construction module, a seabed terrain construction module, a lower-layer motion planning module, an upper-layer motion planning module, and an environmental map output module; the ocean current field construction module is used to construct a three-dimensional ocean current field; the seabed terrain construction module is used to construct a three-dimensional seabed model; the lower-layer motion planning module is used to plan the path between the collection task points and store the generated path information data; the upper-layer motion planning module is used to allocate collection tasks to multiple underwater robots and obtain the target sequence; the environmental map output module is used to draw the target sequence loop diagram.
[0015] The beneficial effects of the present invention are: 1. The present invention conducts task allocation for underwater robots through hierarchical planning, which can efficiently plan the optimal sequence of collection task points under conditions such as a large-scale seabed environment and ocean current interference, realize reasonable and efficient task allocation for multiple robots, and evaluate the states of the allocated robots, effectively solving the problems existing in the current underwater mineral collection process, significantly improving the efficiency of collection tasks, enhancing the comprehensive operation efficiency, and having high robustness and strong practicality.
[0016] 2. The present invention plans the paths between different collection points by using the particle swarm optimization algorithm, and updates the positions and fitness values in the particle swarm optimization algorithm by introducing a three-dimensional ocean current field model, making it more adaptable to the influence of the ocean current environment. The planned paths have better cost and applicability. At the same time, the present invention combines the genetic algorithm and the simulated annealing algorithm to obtain the optimal collection sequence of underwater robots, having higher efficiency in sequence planning, being able to limitedly reduce the energy consumption of the multi-autonomous underwater robot system, and being able to make full and reasonable use of each autonomous underwater robot, having higher use value. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.
[0018] Figure 1 It is the overall flowchart of the mineral collection task scheduling method in the present invention.
[0019] Figure 2 It is a schematic diagram of the seabed terrain model and ocean current field model established in the present invention.
[0020] Figure 3 It is a graph showing the change trend of the lowest total path cost in different iteration times of the present invention.
[0021] Figure 4 It is a schematic diagram of an example of the sequence allocation loop in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0022] The following further illustrates the present invention with reference to the drawings.
[0023] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.
[0024] As Figure 1 shown, a mineral collection task scheduling method for multiple underwater robots includes the following steps: Step 1: Construct a three-dimensional ocean current field model; 1-1. Use the two-dimensional Navier-Stokes equation to construct a viscous Lamb vortex field, and superimpose multiple Lamb vortices to obtain an ocean current field model.
[0025] (1) (2) (3) Among them, , , are the eddy current velocity components in the horizontal, vertical, and vertical axis directions respectively; and are the eddy current intensity and eddy current radius respectively; is the position coordinate vector of the autonomous underwater robot; is the coordinate vector of the eddy current center position; x and y are the position components of the autonomous underwater robot on the horizontal and vertical axes respectively; x 0 and y 0 are the components of the eddy current center coordinates on the horizontal and vertical axes respectively.
[0026] 1-2. Construct an ocean current loss calculation formula, and use the Darcy-Weisbach equation to obtain the pressure loss related to the ocean current in the underwater environment. The formula is as follows: (4) Among them, is the fluid friction loss; is the friction coefficient of seawater, which is affected by the roughness of the fluid surface and the Reynolds coefficient of the fluid; D is the diameter of the autonomous underwater robot; V is the speed of the autonomous underwater robot; is the acceleration due to gravity; is the distance from the current point of the autonomous underwater robot to the next target point, and its expression is: (5) Among them, They are the x, y, and z axis coordinates of the autonomous underwater vehicle at the current moment i respectively.
[0027] Decompose the speed of the autonomous underwater vehicle according to the ocean current direction at the current position to obtain the speed component opposite to the ocean current direction and the speed component consistent with the ocean current direction ; Take the fluid friction loss as the ocean current loss suffered by the speed component and take the negative fluid friction loss as the ocean current loss suffered by the speed component ; During the process of the autonomous underwater vehicle from the current point to the next target point, the ocean current loss is expressed as: (6) Step 2: Construct a three-dimensional seabed terrain model As Figure 2 shown, construct a three-dimensional seabed terrain model. Efficient three-dimensional path planning depends on extracting environmental information from the terrain model. To more accurately describe the three-dimensional seabed environment, construct a three-dimensional seabed terrain model: (7) where is the height at the coordinate ; is the terrain parameter used to control the height of the l-th seabed peak; is the central coordinate of the l-th seabed peak; and are the attenuation factors of the l-th seabed peak along the horizontal axis and vertical axis respectively, used to control the slopes in different directions; is the total number of seabed peaks; .
[0028] Divide the constructed three-dimensional seabed terrain into seamounts, oceanic rises, submarine basins, submarine plateaus, and ridges. Set the coordinates of mineral collection points in the construction of the three-dimensional seabed terrain model according to the Ocean Asset Center database established by the International Seabed Authority (ISA).
[0029] Step 3: Lower layer planning Obtain the optimal path P between different mineral collection points respectively mn ; where and . The method for obtaining the optimal path from mineral collection point m to mineral collection point n is as follows:
[0030] 3-1. Take a path between mineral collection point m and mineral collection point n as a particle in the particle swarm. Set the particle swarm size and the maximum number of iterations, randomly initialize the position and velocity of each particle, and obtain the global optimal position in the initial particle swarm and the individual optimal position of each particle ; where ; D is the particle swarm size.
[0031] 3-2. To adapt to path planning in a three-dimensional environment and make the local planning between lower-level points more adaptable to the ocean current environment, decompose the position update formula and velocity update formula of the traditional particle swarm algorithm, and add the influence of the ocean current at the position where the particle is located. Obtain the velocity and position of the d-th particle after i iterations as follows: (8) (9) where are the velocity components of the d-th particle after i iterations on the x, y, and z axes respectively; w is the inertia weight factor; and are the individual learning factor and the global learning factor respectively; and are random numbers with a value range of 0 to 1; are the individual optimal position components of the d-th particle after i iterations in the x, y, and z directions respectively; are the position components of the d-th particle after i iterations in the x, y, and z directions respectively; are the global optimal position components of the particle after i iterations in the x, y, and z directions respectively.
[0032] 3-3. Perform collision detection on the particle according to the position of the d-th particle after i iterations ; in the three-dimensional seabed terrain model, if the height at the coordinate( ) is greater than or equal to the position component of the d-th particle in the z-axis direction , it is considered that the particle collides with the environment, and the particle is discarded.
[0033] 3-4. Considering the influence of path length and ocean current loss comprehensively, obtain the fitness of each particle after i iterations, and its calculation formula is as follows: (10) where and are the path loss and ocean current loss respectively; and are the path loss weight and ocean current loss weight respectively; They are the x, y, and z axis coordinates of the r-th path point in the particle; R is the number of path points.
[0034] Fitness The smaller it is, the better the corresponding particle; according to the fitness Obtain the global optimal position after the particle swarm iterates i times and the individual optimal position of each particle .
[0035] 3-5. Update the inertia weight factor so that the inertia weight factor decreases as the number of iterations increases, increasing the diversity of the early search and strengthening the local search ability in the later stage to jump out of the local optimal solution. For the individual learning factor and the global learning factor , design an adaptive update method so that it focuses on the individual optimum in the early stage and the global optimum in the later stage, making the particle search range wider and the search speed faster in the early stage, and avoiding falling into the local optimum. The updated factor formula is as follows:
[0036] (11) (12) (13) where and are the maximum and minimum values of the inertia weight factor respectively; is the maximum number of iterations; is the current number of iterations; and are the initial value and the final value of the individual learning factor respectively; and are the initial value and the final value of the global learning factor respectively.
[0037] 3-6. Repeat the above iterative process until the current number of iterations is greater than the maximum number of iterations , and obtain the fitness corresponding to the global optimal position as the path cost Citycost from the mineral collection point m to the mineral collection point n m,n . Construct a path cost matrix Citycost according to the path costs between different mineral collection points, which is expressed as:
[0038] (14) Step Four. Upper-level planning Use the method of random allocation to assign a sequence of mineral collection points to autonomous underwater vehicles. The sequences of mineral collection points corresponding to different autonomous underwater vehicles do not intersect with each other and the sum covers all mineral collection points. Take the optimal collection order in the sequence of mineral collection points corresponding to the autonomous underwater vehicle as the target sequence. The method for obtaining the target sequence of a single autonomous underwater vehicle is as follows: 4-1. Take the arrangement order of different mineral collection points in the sequence of mineral collection points as different individuals in the population to generate an initial population.
[0039] 4-2. Adopt the tournament selection strategy to select multiple individuals from the initial population. Compare the total path costs of the selected individuals, and select the individual with the lowest total path cost as an individual in the parental population. Repeat this process until the number of individuals in the parental population reaches the preset value. The total path cost of an individual is the sum of the path costs between different mineral collection points in the individual.
[0040] 4-3. Randomly select two parental individuals from the parental population and perform crossover using the partially matched crossover method. Respectively obtain the total path costs of the parental individuals and the individuals after crossover. If the total path cost of the individuals after crossover is less than the total path cost of the original parental individuals, then use the individuals after crossover as the offspring individuals; otherwise, use the original parental individuals as the offspring individuals. Repeat this process until the size of the offspring population reaches the preset value.
[0041] 4-4. Randomly select offspring individuals from the offspring population for mutation operations, randomly adjust the order of mineral collection points in the selected offspring individuals. If the total path cost of the individuals after mutation is less than the total path cost of the original offspring individuals, then replace the original offspring individuals and add them to the offspring population.
[0042] 4-5. Simulated annealing Randomly select an offspring individual from the offspring population and use the 2-opt (two-element optimization) method to transform the path of the offspring individual. If the total path cost of the individual after transformation is less than the total path cost of the original offspring individual, then replace the original offspring individual with the individual after transformation and add it to the offspring population; otherwise, with probability Replace the original offspring individual with the individual after transformation and add it to the offspring population; where is the difference between the total path costs of the individual after transformation and the original offspring individual; is the current temperature. Loop this process until the loop count k is greater than the maximum loop count K. The current temperature in each loop process .
[0043] 4-6. Continuously iterate the processes of selection, crossover, mutation, and simulated annealing. Regenerate the initial population in each iteration process until the iteration count j reaches the maximum iteration count J. Select the individual with the lowest total path cost among all iteration counts as the target sequence. The change trend of the lowest total path cost in different iteration counts is asFigure 3 as shown
[0044] Step Five: Obtain the target sequences of all autonomous underwater vehicles respectively, and complete the mineral collection task scheduling for the autonomous underwater vehicles according to the target sequences, as Figure 4 shown
[0045] The mineral collection task scheduling device adopted by the mineral collection task scheduling method includes an ocean current field construction module, a seabed terrain construction module, a lower-layer motion planning module, an upper-layer motion planning module, and an environmental map output module; the ocean current field construction module is used to construct a three-dimensional ocean current field; the seabed terrain construction module is used to construct a three-dimensional seabed model; the lower-layer motion planning module is used to plan the paths between the collection task points and store the generated path information data; the upper-layer motion planning module is used to allocate collection tasks to multiple underwater vehicles and obtain the target sequences to ensure the efficiency and rationality of task allocation; the environmental map output module is used to draw the target sequence loop diagram and provide a visual final task allocation effect.
Claims
1. A method for scheduling mineral collection tasks for multiple underwater robots, characterized in that: The following steps are involved: Step 1: Construct a three-dimensional ocean current field model; Step 2: Obtain a three-dimensional seabed terrain model; use the particle swarm algorithm to obtain the optimal paths between different mineral collection points in the three-dimensional seabed terrain model. In each iteration, introduce the constructed three-dimensional ocean current field model to update the position and fitness of each particle; Step 3: Assign a sequence of mineral collection points to the underwater robot; use the optimal collection order in the mineral collection point sequence corresponding to the underwater robot as the target sequence, obtain the target sequences of all underwater robots according to the optimal paths between different mineral collection points, and use the target sequence to complete the mineral collection task scheduling of the underwater robot.
2. The method for scheduling mineral collection tasks for multiple underwater robots according to claim 1, characterized in that: In the step 2, the method for updating the position of each particle is as follows: in, are the position components of the d-th particle in the x, y, and z directions after iteration i; are the velocity components of the d-th particle on the x, y, and z axes after iteration i; , , are the eddy current velocity components in the horizontal, longitudinal and vertical directions respectively; ; D is the particle group size; is the current iteration number.
3. The method for scheduling mineral collection tasks for multiple underwater robots according to claim 1, characterized in that: In the step 2, the method for updating the fitness of each particle is as follows: in, is the fitness of the particle; and are path loss and ocean current loss respectively; and are path loss weight and ocean current loss weight respectively; are the x, y, and z axis coordinates of the rth path point in the particle, respectively; R is the number of path points; is the friction coefficient of seawater; is the distance from the current path point to the next path point of the underwater robot; V is the speed of the underwater robot; D is the diameter of the underwater robot; is the acceleration due to gravity.
4. The method for scheduling mineral collection tasks for multiple underwater robots according to claim 1, characterized in that: In the step 2, the iterative update method of the individual learning factor and the global learning factor in the particle swarm algorithm is as follows: in, is the individual learning factor; is the global learning factor; and are the initial and final values of the individual learning factor, respectively; and are the initial and final values of the global learning factor, respectively; is the current iteration number; is the maximum number of iterations.
5. The method for scheduling mineral collection tasks for multiple underwater robots according to claim 1, characterized in that: In the step 3, the method for obtaining the target sequence of the underwater robot is: using different arrangement orders of the mineral collection points in the collection point sequence as the initial population; continuously iterating the process of selection, crossover, mutation, and simulated annealing, and regenerating the initial population before each iteration until the number of iterations reaches the maximum number of iterations; The individual with the lowest total path cost among all iterations is selected as the target sequence.
6. The method for scheduling mineral collection tasks for multiple underwater robots according to claim 5, characterized in that: The method for obtaining the total path cost is as follows: taking the fitness corresponding to the optimal path between different mineral collection points as its path cost, and adding the path costs between different mineral collection points in sequence according to the arrangement order of the mineral collection points in the collection point sequence to obtain the total path cost.
7. The method for scheduling mineral collection tasks for multiple underwater robots according to claim 6, characterized in that: The method for simulating degradation is as follows: Randomly select individuals from the population obtained after mutation, and use the two-element optimization method to transform the individual's path; if the total path cost of the transformed individual is less than the total path cost of the original individual, the transformed individual replaces the original individual and joins the population; otherwise, the probability The transformed individuals replace the original individuals and join the population; among them, is the difference between the total path cost of the transformed individual and the original individual; is the current temperature; the process is repeated until the number of cycles k is greater than the maximum number of cycles K.
8. The method for scheduling mineral collection tasks for multiple underwater robots according to claim 1, characterized in that: In the step 1, the method for constructing a three-dimensional ocean current field model is: using the Navier-Stokes equations to construct a viscous Lamb vortex field, and superimposing multiple Lamb vortices to obtain an ocean current field model.
9. The method for scheduling mineral collection tasks for multiple underwater robots according to claim 1, characterized in that: In the step three, the mineral collection point sequences corresponding to different underwater robots do not intersect with each other and the sum covers all the mineral collection points.
10. A mineral collection task scheduling system for multiple underwater robots, characterized in that: Used to execute the mineral collection task scheduling method for multiple underwater robots as described in claim 1; the mineral collection task scheduling system includes an ocean current field construction module, a seabed terrain construction module, a lower-level motion planning module, an upper-level motion planning module and an environment map output module; The ocean current field construction module is used to construct a three-dimensional ocean current field; the seabed terrain construction module is used to construct a three-dimensional seabed model; the lower-level motion planning module is used to plan the paths between the collection task points and store the generated path information data; the upper-level motion planning module is used to assign collection tasks to multiple underwater robots and obtain target sequences; the environment map output module is used to draw the target sequence loop diagram.