Amphibious unmanned vehicle water area shape and motion control synchronous optimization design method

Through the morphological intelligent double-layer optimization method, the water appearance and motion control strategies of amphibious unmanned vehicles are synchronously optimized, which solves the problem of separate design in the existing technology and achieves automated design and performance improvement.

CN118916991APending Publication Date: 2024-11-08NAT INNOVATION INST OF DEFENSE TECH PLA ACAD OF MILITARY SCI
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410976664.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-21
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

The existing amphibious unmanned vehicle water appearance design and motion control strategies are designed separately, relying on manual experience, lacking automated design methods, and do not involve synchronous optimization of appearance design and motion control strategies.

Method used

Using a two-layer optimization method based on morphological intelligence, the simulation environment is constructed, combined with morphological space, environmental interaction, control strategies and evaluation criteria, and synchronous optimization design of amphibious unmanned vehicle water appearance and motion control strategies is realized, and reinforcement learning and genetic algorithms are used. Iterative optimization is carried out to obtain the optimal appearance and motion control strategy.

Benefits of technology

The automation of amphibious unmanned vehicle design has been realized, which reduces the dependence on experienced designers, improves design efficiency and performance, ensures the best combination of appearance and motion control strategies, and improves overall performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118916991A_ABST
    Figure CN118916991A_ABST
Patent Text Reader

Abstract

The invention discloses a synchronous optimization design method for water area appearance and motion control of an amphibious unmanned vehicle, and belongs to the technical field of software and hardware optimization design of amphibious vehicles. According to the method, four elements of form intelligent form space, environment interaction, a control strategy and an evaluation criterion are utilized to construct a form space, form variables, a change range and a water area motion control strategy of the water area appearance of the amphibious unmanned vehicle, evaluate the evaluation criterion of the water area expression of the amphibious unmanned vehicle, construct an inner ring and an outer ring, and improve the accuracy of the amphibious unmanned vehicle. The shape and control strategy of the amphibious unmanned vehicle are continuously optimized and evolved, an optimal result is obtained, and rapid research, development and use of amphibious unmanned vehicles with different purposes are facilitated. The amphibious unmanned vehicle water area appearance is obtained through rapid optimization iteration by utilizing a form intelligence method according to the performance indexes pulled by task requirements, and a designer with rich experience can update or adjust the scheme on the basis of the result, so that the dependence on manpower is greatly reduced, and even the link of manual design can be completely replaced subsequently.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of software and hardware optimization design of amphibious vehicles, and specifically is a method for synchronously considering and optimizing the design of the water body shape and motion control strategy of an amphibious unmanned vehicle. Background Art

[0002] Amphibious vehicles have the functions of driving on the ground and on the water. They are widely used in the civilian field for flood control, disaster relief, seaside patrol, material transportation in wading areas, beach entertainment, etc. Due to the different use focuses, amphibious vehicles have very different designs in terms of appearance and running devices. There are many technical routes and great differences, which poses a severe challenge to the design ability of designers. For example, when amphibious vehicles are used for flood control and disaster relief, they are required to be small and flexible, and able to shuttle freely in rapids and ruins; when amphibious vehicles are used for material transportation, they are required to have a large load capacity and a high load-to-weight ratio; when amphibious vehicles are used for border and sea patrol, they are required to be fast, flexible in switching between land and water, and able to catch up and deal with the enemy in time when they are found. In addition to different tasks, the use environment of amphibious vehicles will also affect their design plans. For example, when amphibious vehicles are mainly used in water environments, their appearance tends to be more like a ship, while when amphibious vehicles are mainly used in land environments, their appearance tends to be more like a car model.

[0003] Different uses of amphibious vehicles require different performance and functions of the platform, resulting in different forms and motion control strategies for amphibious vehicles. How to design an amphibious unmanned vehicle? The current mainstream technology relies entirely on designers to complete it based on specific needs and manual experience.

[0004] The design of a general amphibious unmanned vehicle follows the following steps: investigate needs and analyze the current state of technology, set design goals according to tasks, draft design sketches, build three-dimensional models, make and test models, optimize designs according to model tests, make and debug physical objects, debug autonomous related technologies such as motion control, and so on. During this period, the drafting of the design sketch of the amphibious unmanned vehicle was mainly completed by experienced designers with the assistance of design tools. For example, patent publication number CN206426775U proposes a design method for a lightweight, high-speed wheeled amphibious vehicle body, and the appearance adopts a streamlined arc shape to reduce resistance during movement in water. Patent publication number CN111559211A proposes a water drag reduction mechanism in order not to destroy the assembly line structure of the amphibious vehicle appearance, using a retractable skateboard to fill the incomplete part of the hydrodynamic streamline in the amphibious vehicle appearance. Patent application 202110913632.7 and patent publication number CN212353504U both propose design methods for walking mechanisms, but the above design schemes are all manually designed by experienced designers.

[0005] How to use new methods to assist or replace experienced designers to achieve automated product design has always been a key research direction in the field of scientific research. For example, Patent Publication No. CN113032902A uses machine learning methods to optimize the design of the aerodynamic head shape of high-speed trains, Patent Publication No. CN116432300A uses manifold learning methods to achieve the design of underwater glider shapes, and Patent Publication No. CN114676639A uses neural network learning architecture to achieve the optimal design of aircraft aerodynamic shapes. These are all attempts to use machine learning methods to achieve the optimal design of product shapes and reduce the workload of designers.

[0006] Currently, there is no literature on the use of machine learning methods to achieve automatic optimization design of the water shape of amphibious unmanned vehicles. In addition, there is no literature on the research of coupling the shape design and the subsequent motion control strategy design to achieve synchronous optimization design. Summary of the invention

[0007] (I) Purpose of the invention

[0008] The purpose of the present invention is to provide a synchronous optimization design method for the land walking device and motion control of an amphibious unmanned vehicle. The technical problem to be solved by the present invention is that the existing water area appearance design and water area motion control strategy of an amphibious unmanned vehicle are designed separately and independently, belonging to two different steps, and both are currently basically designed and optimized based on artificial experience. The present invention provides a synchronous optimization design method for the water area appearance and motion control strategy of an amphibious unmanned vehicle based on morphological intelligence. The method is based on the idea of ​​internal and external double-layer optimization of morphological intelligence, and adopts four elements of morphological space, environmental interaction, control strategy, and evaluation criteria as means. The constructed simulation environment is used to realize the performance evaluation of the water area appearance and motion control strategy of the amphibious unmanned vehicle, and the optimization and iteration are continuously performed according to the evaluation results to obtain the optimal appearance and optimal motion control strategy under specific performance indicators, thereby replacing the traditional design method that relies on artificial intelligence.

[0009] (II) Technical solution

[0010] In order to achieve the above objectives and solve the above technical problems, the present invention adopts the following technical solutions:

[0011] The synchronous optimization design method of the water body shape and motion control of an amphibious unmanned vehicle includes the following steps:

[0012] Step 1: Determine the key performance indicators of the amphibious unmanned vehicle according to its specific purpose, such as the expected length, width and height, the maximum speed in the expected water area, the maximum load, the draft, etc.;

[0013] Step 2: According to the expected length, width and height parameters, construct the morphological space of the amphibious unmanned vehicle, set the spatial variables and the range of variation. The morphological space of the amphibious unmanned vehicle is relatively complex, and can be simplified into a cuboid, trapezoid, quadrangular pyramid and other main parts, and then use the length, width, height, inclination and sideline of these parts as spatial variable parameters to achieve the morphological change of the shape;

[0014] Step 3: Construct a simulation interactive environment for optimizing the appearance of an amphibious unmanned vehicle in water. The simulation interactive environment needs to establish a dynamic model and a kinematic model of the amphibious unmanned vehicle on water, simulate the movement of the amphibious unmanned vehicle in water under various water conditions, simulate the effects of water resistance, buoyancy, waves, etc. on the vehicle body, and provide corresponding data.

[0015] Step 4: Construct the input and output variables and constraints of the motion control strategy of the amphibious unmanned vehicle in water. The motion control input variables of the amphibious unmanned vehicle driving in water generally consist of the current vehicle's throttle, direction, vehicle body posture, draft, current position water depth status information and the "driver's" expected speed, expected path and other expected control information. The throttle and direction in water are mainly reflected in two sets of hydraulic cylinders that control the size of the spray pump and the direction of the spray pump. The "driver" here can be a remote control operator or a computer program for automatic driving. The motion control output variables of the amphibious unmanned vehicle driving in water include throttle, brake, and direction information, which are sent to the vehicle chassis to control the real movement of the amphibious unmanned vehicle.

[0016] Step 5: Set up the amphibious unmanned vehicle water motion simulation task; the task in the water can be simply set to move from the starting point to the end point, or some floating barrels and other ships can be arranged between the starting point and the end point to simulate obstacles and test the amphibious unmanned vehicle's ability to bypass obstacles;

[0017] Step 6: Construct an evaluation criterion system for the water body shape and motion control strategy of the amphibious unmanned vehicle; when the amphibious unmanned vehicle is driving in water to complete a specific task, the evaluation criteria can be task completion (such as whether it can reach the end point), maximum driving speed, average driving speed, task completion status, maximum load of the completed task, safe distance from obstacles, vehicle body bumpiness (stability), etc., and one or more combinations can be used for measurement according to actual task requirements;

[0018] Step 7: Construct a two-layer optimization framework for the water shape and motion control strategy of the amphibious unmanned vehicle, and select a suitable reinforcement learning algorithm, that is, use the morphological intelligence method to optimize the motion control strategy using the reinforcement learning algorithm in the inner layer and the shape using the genetic algorithm in the outer layer;

[0019] Step 8: Construct a mapping relationship table between the morphological space under different parameters and the score in the simulation interaction environment; the simulation interaction of the water area shape involves computational fluid dynamics, water area grid generation and other calculations. The simulation software requires high computing resources and is time-consuming. In addition, there are many parameters used to represent the shape. Therefore, it is necessary to generate the mapping relationship between the morphological space under different parameters and the score in the simulation interaction environment in advance;

[0020] Step 9: Put the initial shape constructed in the second step and the motion control strategy constructed in the fourth step into the double-layer optimization framework. Using the simulation interaction environment constructed in the third step, the amphibious unmanned vehicle performs the water motion simulation task constructed in the fifth step. The evaluation criteria system constructed in the sixth step is used to evaluate the progress of the amphibious unmanned vehicle in executing the task, and the evaluation criteria are ranked according to the evaluation level. The specific double-layer cycle is as follows:

[0021] Step 9.1: For the input shape set and the matching motion control strategy, execute the set task in the simulation interactive environment, evaluate the task completion, and obtain the evaluation score;

[0022] Step 9.2: Perform inner loop optimization, which mainly optimizes and iterates the motion control strategy based on reinforcement learning. When the shape remains unchanged, the motion control strategy is updated, and the same type of task is performed in the same simulation interactive environment with the new motion control strategy. The task completion is evaluated (search the score mapping table in step 8), and a new score is obtained and added to the score ranking list;

[0023] Step 9.3: Repeat step 9.2 to train the reinforcement learning strategy until its motion control strategy under the shape condition is optimal, that is, the score is the highest;

[0024] Step 9.4: Perform morphological evolution on the top-scoring shapes (with corresponding motion control strategies) in the above inner loop, i.e., use the morphological variables to "mutate" within the preset range of variation, and evolve according to the variation rules of the morphological variable parameters constructed in advance to obtain several groups of new shapes, forming a new generation of shape sets;

[0025] Step 9.5: If the comprehensive score on the ranking list does not continue to improve, or the number of iterations meets the requirement, it is considered that the optimal score has been reached, and then go to step 10, otherwise go to step 9.2;

[0026] Step 10: Get the best result, that is, the most suitable shape and matching motion control strategy, for experienced designers to confirm or optimize based on this.

[0027] (III) Effective income

[0028] Compared with the prior art, the advantages of the present invention are:

[0029] 1. The present invention can be used to quickly optimize and iterate the water shape of the amphibious unmanned vehicle based on the performance indicators towed by the mission requirements using the morphological intelligence method. Experienced designers can update or adjust the plan based on the results, thereby greatly reducing the dependence on experienced designers, and even completely replacing the manual design link in the future;

[0030] 2. The present invention considers the motion control strategy of the amphibious unmanned vehicle in water after it is built and used while considering the appearance design in the design stage of the amphibious unmanned vehicle. In theory, the motion control strategy is closely related to the operating performance of the appearance in water. Only a combination of different appearances and adaptive control strategies can bring out the overall optimal performance. Therefore, the appearance and the motion control strategy are optimized simultaneously in the design stage, that is, the optimized design that couples the shape and the action together can achieve better overall performance than a separate independent design. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a schematic diagram of the process of the present invention;

[0032] Figure 2 It is a schematic diagram of the double-layer optimization iterative framework of the present invention;

[0033] Figure 3 It is a spatial diagram of the ship's external form, which is illustrated by way of example;

[0034] Figure 4 It is the morphological variables and range of variation of the ship’s shape;

[0035] Figure 5 It is a schematic diagram of the state of the ship's shape under different variable parameters in the morphological space;

[0036] Figure 6 and Figure 7 This is the working interface diagram of the water area simulation software STAR-CCM. DETAILED DESCRIPTION

[0037] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0038] The present invention proposes to use a morphological intelligence method to achieve a synchronous optimization design method for the appearance and motion control of an amphibious unmanned vehicle. Unlike the traditional appearance design and motion control debugging considered separately, the present invention believes that the optimization design coupled together is better than the results obtained by independent design optimization. Because the present invention believes that the motion control strategy is related to the appearance, the optimal motion control strategy corresponding to different appearances is different. At present, these two links are considered and designed separately, and the motion strategy is designed after the appearance is determined. The effect is not as good as the design that considers the two parts simultaneously.

[0039] The morphological intelligence method is inspired by nature, and believes that the development of intelligence is not only related to one's own brain and body, but also closely related to the environment in which it is located. The morphological intelligence theory believes that the survival of the fittest in nature is passed down from generation to generation, which only transmits the shape of the species, while the various survival skills of the species need to be learned and trained in the future. However, robots are different from biological organisms in nature. In the process of iterative updates, the offspring of robots can not only optimize the appearance based on the inheritance of the parent generation, but also directly inherit the control strategy from the parent generation and optimize it on this basis without having to start learning again. Based on this theory, the literature Embodied Intelligence via Learning and Evolution believes that in order to adapt to tasks and environments, intelligent agents need to optimize their body shape and brain control strategies simultaneously, and cooperate with each other to evolve better results than evolution alone, that is, the concept of achieving better results through dual optimization of the inner and outer layers of the intelligent body. At present, the morphological intelligence method is still in the stage of theoretical exploration and simple simulation verification, and no literature related to the use of morphological intelligence to simultaneously optimize the design of product structures and product control strategies such as robots has been found.

[0040] The present invention proposes a method for realizing the synchronous optimization design of the appearance and motion control strategy of amphibious unmanned vehicles in water areas by using morphological intelligence methods, which is used to replace the two steps of designers designing their appearance according to specific tasks and debugging motion control strategies separately. The four elements of morphological intelligence morphological space, environmental interaction, control strategy, and evaluation criteria are used to construct the morphological space of the appearance of amphibious unmanned vehicles in water areas, morphological variables and range of variation, water area motion control strategy, and evaluation criteria for evaluating the performance of amphibious unmanned vehicles in water areas. The inner and outer loops are constructed to realize the continuous optimization and evolution of the appearance and control strategy of amphibious unmanned vehicles, obtain the best results, and help the rapid development and use of amphibious unmanned vehicles for different purposes.

[0041] Example 1

[0042] like Figure 1 As shown, the steps of the synchronous optimization design method for the appearance and motion control of an amphibious unmanned vehicle proposed by the present invention are as follows:

[0043] Step 1: Determine the expected design indicators based on the expected performance and functions of the amphibious unmanned vehicle. For example, taking a small amphibious unmanned vehicle mainly used for rapid transportation as an example (the same below), its expected indicators are: length, width and height ≤ 800cm × 300cm × 250cm, weight ≤ 4000kg, load ≥ 300kg, maximum speed on water ≥ 40km / h, wind and wave resistance ≥ Level 3 sea conditions, etc.;

[0044] Step 2: According to the task of optimizing the water area shape, construct its morphological space, morphological variables and variation range. Figure 3As shown in the figure, the shape of the ship is expressed in a three-stage mathematical way. The shape can be represented by lines and points, and several parameters (six parameters of the front part of the vehicle are used in this example) are taken as morphological variables. The range of variable changes is configured according to artificial experience and mutual constraints, as shown in the figure. Figure 4 As shown;

[0045] Step 3: Select a suitable simulation interaction environment for simulating the model. This example uses the water simulation software STAR-CCM as the simulation interaction environment. Figure 6 and Figure 7 The working interface diagram of the STAR-CCM simulation software for optimizing the shape is shown. The simulation software is used to automatically calculate the various performances of unmanned vehicles with different shapes moving in water.

[0046] Step 4: Set the input and output variables and constraints of the motion control strategy. This example is to optimize the shape and motion control strategy of an amphibious unmanned vehicle during its driving in water. Therefore, for the optimization goal, the simulation input parameters include the position information, attitude information, speed information, direction information, water depth, speed and direction information expected by the "driver", etc. of the unmanned vehicle. The output control strategy parameters include the expected speed and expected direction for controlling the vehicle's motion.

[0047] Step 5: Set the simulation task. In this example, the motion task that requires the car to run from point A to point B is set in the simulation interactive environment.

[0048] Step 6: Set the evaluation criteria. In this example, the evaluation criteria include water movement speed, total time, stability during movement, load capacity, etc. You can set one or a combination of these to evaluate the completion of the simulation task;

[0049] Step 7: Select the inner and outer optimization algorithms. The framework of the inner and outer optimization algorithms is as follows: Figure 2 As shown, in this example, the inner layer selects the Promimal policy optimization (PPO) algorithm in reinforcement learning, and the outer layer selects the genetic algorithm;

[0050] Step 8: Construct a mapping relationship table between the morphological space under different parameters and the score in the simulation interaction environment; first, the water simulation software STAR-CCM involves fluid mechanics simulation, which requires powerful computing resources and a lot of computing time, and no one can achieve real-time simulation; secondly, the 6 parameters (X1, X2, Y1, Y2, Y3, Y4) of the front part of the vehicle in this example can all be randomly selected within a preset range of variation, so it is a parameter space of 6-dimensional continuous variables, and the traversal space is complex, and the amount of calculation required after the combination is huge, and in the actual application process, in order to achieve a better design effect, the variable parameters of the shape will only be set more, resulting in a greater amount of calculation required. Therefore, the present invention uses a machine learning method to first train a mapping relationship table between the morphological space under different parameters and the score in the simulation interaction environment, as follows:

[0051] Step 8-1: Evenly select n values ​​within the range of each parameter (n is generally greater than 2 and less than or equal to 5, and n=3 is taken in this example), and then get n 6 Group data (3 in this example 6 =729) as training samples;

[0052] Step 8-2: Use this n 6 The appearance of the group data is simulated and run in a simulation interactive environment to obtain the corresponding score;

[0053] Step 8-3: Use machine learning to train the above training samples to obtain a mapping relationship table between the morphological space under different parameters and the score in the simulation interaction environment.

[0054] Step 9: Perform double-layer optimization iteration;

[0055] Step 9-1: Evaluate the input shape set and the matching motion control strategy in a simulated interactive environment to obtain an evaluation score;

[0056] Step 9-2: Perform inner loop optimization, and evaluate the new motion control strategy in the water simulation interactive environment according to the reinforcement learning strategy to obtain a new score;

[0057] Step 9-3: Repeat step 9-2 to train the reinforcement learning strategy until its motion control strategy under the shape condition is optimal, that is, the score is the highest (the score no longer continues to increase and the number of cycles reaches N times, where N is the minimum number of cycles set manually);

[0058] Step 9-4: Perform morphological evolution on the above-mentioned top-scoring shapes (with corresponding motion control strategies), that is, use the morphological variables to "mutate" within a preset range of variation to obtain several groups of new shapes to form a new generation of shape sets; Figure 5 Shows some of the morphological forms in the evolutionary process of the morphological space.

[0059] Step 9-5: If the comprehensive score does not continue to improve and the number of cycles reaches M (M is the minimum number of cycles set manually), go to step 10, otherwise go to step 9-2;

[0060] Step 10: Output the shape with the highest comprehensive score and the corresponding motion control strategy as the optimal design result obtained by this method for confirmation by experienced designers.

[0061] Through the above steps, the optimal water shape and motion control strategy design of the amphibious unmanned vehicle can be obtained.

[0062] This method is guided by the theory of morphological intelligence and uses the morphological intelligence double-layer optimization framework and the four-element optimization evolution idea to design the optimal water shape and water motion control strategy of an amphibious unmanned vehicle. It is significantly different from traditional reinforcement learning methods that only optimize control strategies or traditional genetic algorithms that only inherit and mutate the shape. It avoids the drawbacks of separate consideration and design of shape and motion control strategies in traditional methods, and improves design performance and efficiency.

[0063] The examples given in this method are only for better describing the process of the method of the present invention. Its optimization objectives, morphological space, evaluation criteria, constructed simulation interaction environment, adopted double-layer optimization algorithm, etc. are not limited to the algorithms mentioned in the examples. The number and range of morphological variables, the number of evaluation criteria, the number of optimization iterations, etc. are configured by designers according to the specific optimization tasks, but do not affect the method process proposed in the present invention.

[0064] It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications without departing from the principle of the present invention should be regarded as within the scope of protection of the present invention.

[0065] The above contents are further detailed descriptions of the present invention in combination with specific implementation methods, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the protection scope of the present invention.

Claims

1. A method for synchronous optimization design of water body shape and motion control of amphibious unmanned vehicle, characterized in that: The steps include: Step 1: Determine the key performance indicators of the amphibious unmanned vehicle according to the specific purpose of the actual mission, including the expected length, width and height, the maximum speed in the expected water area, the maximum load, and the draft depth; Step 2: According to the expected length, width and height parameters, construct the amphibious unmanned vehicle shape space, set the spatial variables and the range of variation; the shape space of the amphibious unmanned vehicle is composed of a cuboid, a trapezoid and a quadrangular pyramid, and select a part of the length, width, height, inclination and edge of the above components as the spatial variable parameters, so as to realize the shape change of the shape; Step 3: Construct a simulation interactive environment for optimizing the appearance of an amphibious unmanned vehicle in water. The simulation interactive environment needs to establish a kinematic model of the amphibious unmanned vehicle on water, simulate the movement of the amphibious unmanned vehicle in water under various water conditions, simulate the effects of water resistance, buoyancy, and waves on the vehicle body, and provide corresponding data. Step 4: Construct the input and output variables and constraints of the amphibious unmanned vehicle's water motion control strategy; The motion control input variables of an amphibious unmanned vehicle driving in water are composed of the current vehicle's throttle, direction, vehicle posture, draft, current water depth status information, and the "driver's" desired speed and desired path. The motion control output variables of the amphibious unmanned vehicle driving in water include throttle, brake, and direction information, which are sent to the vehicle chassis to control the actual motion of the amphibious unmanned vehicle; Step 5: Set up a water motion simulation task for the amphibious unmanned vehicle; set the task in the water to move from the starting point to the end point, or arrange several floating barrels and other ships between the starting point and the end point to simulate obstacles, so as to test the obstacle avoidance ability of the amphibious unmanned vehicle; Step 6: Construct an evaluation criterion system for the water body shape and motion control strategy of the amphibious unmanned vehicle; When an amphibious unmanned vehicle is driving in water to complete a specific task, the evaluation criteria include: task completion, maximum driving speed, average driving speed, task completion status, maximum load of the completed task, safe distance from obstacles, and vehicle body bumpiness. One or more combinations are used for measurement according to actual task requirements. Step 7: Construct a two-layer optimization framework for the water shape and motion control strategy of the amphibious unmanned vehicle, and select a suitable reinforcement learning algorithm, that is, use the morphological intelligence method to optimize the motion control strategy using the reinforcement learning algorithm in the inner layer and the shape using the genetic algorithm in the outer layer; Step 8: Put the initial shape constructed in step 2 and the motion control strategy constructed in step 4 into the double-layer optimization framework. Using the simulation interaction environment constructed in step 3, the amphibious unmanned vehicle performs the water motion simulation task constructed in step 5. Use the evaluation criteria system constructed in step 6 to evaluate the progress of the amphibious unmanned vehicle in performing the task, and sort them according to the evaluation level. The specific double-layer cycle is as follows: Step 8.1: For the input shape set and the matching motion control strategy, execute the set task in the simulation interactive environment, evaluate the task completion, and obtain the evaluation score; Step 8.2: Perform inner loop optimization, mainly optimizing and iterating the motion control strategy based on reinforcement learning. When the shape remains unchanged, update the motion control strategy, use the new motion control strategy to perform the same type of task in the same simulation interactive environment, evaluate the task completion, obtain a new score, and add it to the score ranking list; Step 8.3: Repeat step 8.2 to train the reinforcement learning strategy until the updated motion control strategy is optimal under the shape condition, that is, the score is the highest; Step 8.4: Perform morphological evolution on the top-scoring shapes in the inner loop (with the corresponding motion control strategies), that is, use the morphological variables to "mutate" within the preset range of variation, and evolve according to the variation rules of the morphological variable parameters constructed in advance to obtain several groups of new shapes, forming a new generation of shape sets; Step 8.5: If the comprehensive score on the ranking list does not continue to improve, or the number of iterations meets the requirement, it is considered that the optimal score has been reached, and then go to step 9, otherwise go to step 8.2; Step 9: Get the optimal shape and matching motion control strategy for the designer to confirm or optimize based on it.

2. The method for synchronous optimization design of water area shape and motion control of an amphibious unmanned vehicle according to claim 1 is characterized in that: In step 4, the throttle and direction are mainly reflected in the water as two groups of hydraulic cylinders that control the size of the spray pump and the direction of the spray pump's water spraying.

3. The method for synchronous optimization design of water area shape and motion control of an amphibious unmanned vehicle according to claim 1 is characterized in that: In step 4, the "driver" is a remote control operator or a computer program for automatic driving.

4. The method for synchronous optimization design of water area shape and motion control of an amphibious unmanned vehicle according to claim 1 is characterized in that: In step 4, the "driver" is a remote control operator or a computer program for automatic driving.

Citation Information

Cited By

  • Task-oriented resource allocation scheme optimization method for amphibious unmanned vehicle

    CN119831202A