Robot distributed control method and system based on hydrodynamic principle

Abstract

The invention discloses a robot distributed control method and system based on a fluid dynamics principle, and the method comprises the steps: obtaining the image information of a target region, extracting the contour of a target pattern, and constructing a truncated signed distance field to generate target guide potential energy; the robot is uniformly divided into a plurality of sub-regions for initial deployment; all the robots are modeled as particle elements, and interaction force between the robots is constructed through a smoothed particle fluid mechanics method; a distributed control strategy is designed for each robot, the control input of the distributed control strategy is the sum of target guiding force and interaction force, and meanwhile damping control is combined to stabilize system response; in the control process, as the robots enter the target area, the target guiding force is gradually attenuated to zero, only the interaction force between the robots is reserved, and therefore automatic construction of the target pattern is achieved. According to the invention, a completely distributed pattern construction and space regulation and control strategy is realized, and the method has good expansibility, adaptive ability and control stability.

Description

Technical Field

[0001] This invention relates to the field of swarm robot systems and distributed control technology, and in particular to a robot distributed control method and system based on the principles of fluid dynamics. Background Technology

[0002] With the rapid development of artificial intelligence and intelligent robotics, swarm robot systems have received widespread attention in fields such as environmental monitoring, disaster search and rescue, collaborative construction, and unmanned patrolling. Traditional swarm control methods typically rely on predefined paths, central schedulers, or global task allocation mechanisms. While these methods perform well in structured environments, they are prone to problems such as high computational complexity, poor adaptability, and insufficient scalability when facing dynamically changing or complex-shaped target areas.

[0003] In recent years, distributed control strategies have become an important research direction in swarm robotic systems. These methods emphasize collaborative control among robots through local perception and short-range communication, thereby improving the system's adaptability in unknown, dynamic, or communication-constrained environments. However, existing distributed methods still face many challenges in practical applications, such as the inability to guarantee collision avoidance between robots, difficulty in maintaining pattern consistency in complex or non-convex regions, slow shape construction speed, and uneven spatial coverage.

[0004] Physics-inspired control models, especially those based on fluid dynamics, offer new insights into solving these problems. Fluid dynamics possesses excellent local consistency, self-organization capabilities, and distributed coordination characteristics. Introducing it into the control of swarm robot systems holds promise for achieving collision-free, uniform, and adaptive configuration control within different target regions. Therefore, designing a fluid dynamics-based distributed control method that enables swarm robots to achieve high-fidelity pattern generation and flexible deployment in dynamic environments remains a pressing technical challenge. Summary of the Invention

[0005] The purpose of this invention is to provide a robot distributed control method and system based on the principle of fluid dynamics, which realizes a fully distributed pattern construction and spatial regulation control strategy, avoids dependence on global scheduling and spatial discretization, and has good scalability, adaptability and control stability.

[0006] To achieve the above objectives, the present invention provides the following solution:

[0007] A distributed control method for robots based on fluid dynamics principles includes the following steps:

[0008] S1. Obtain image information of the target area, extract the target pattern contour, construct a truncated signed distance field to generate the target guiding potential energy, and calculate the target guiding force;

[0009] S2. Based on the area of ​​the target pattern, construct the central gathering area and evenly divide the robot into multiple sub-areas for initial deployment;

[0010] S3. Model each robot as a particle volume element with local perception and second-order dynamic characteristics, and construct the interaction forces between robots through the smooth particle fluid dynamics method, including pressure force, viscous force and density estimation.

[0011] S4. Design a distributed control strategy for each robot, with the control input being the sum of the target guiding force and the interaction force, while combining damping control to stabilize the system response.

[0012] S5. During the control process, as the robot enters the target area, the target guiding force gradually decays to zero, and only the interaction force between the robots is retained to maintain the spatial uniformity of the configuration and the stability of the pattern boundary, thereby realizing the automatic construction of the target pattern.

[0013] Preferably, in S1, the target guiding force is calculated from the gradient direction of the truncated signed range field corresponding to the target pattern, and has the ability to attract the robot into the target area. The calculation formula for the target guiding force is as follows:

[0014]

[0015] in, It is the target attraction gain. The target guiding force intensity factor, Guiding force for the goal, Let represent the unsigned distance field to the boundary of the target region.

[0016] Preferably, in S3, the interaction forces between robots are weighted using a two-dimensional cubic spline kernel function. Satisfies normality and compact support:

[0017]

[0018] in, It is the normalized distance between robots. To smooth out the length, the kernel assigns higher weights to nearby robots.

[0019] In the preferred color scheme, S3, the interaction forces between robots are constructed using a smooth particle hydrodynamics method, specifically including:

[0020] robot density Defined as:

[0021]

[0022] in, For kernel function, For robots Location, For robots Location; pressure corresponding to density Using the Tate state equations, it can be expressed as:

[0023]

[0024] in, For reference density, The stiffness index is... It is a pressure scaling factor related to local density; robot and viscosity terms between use Viscous form:

[0025]

[0026] in, Represents robots and The relative speed between them Represents robots and The relative positions between them , The viscosity coefficient, Represents robots and average density, Denotes their average speed of sound, where ,item As an approximation of the velocity divergence term, it is defined as: ;

[0027] Actions on robots Total interaction force based on smooth particle hydrodynamics The sum of the pressure and viscous forces exerted by all its neighboring robots:

[0028]

[0029] in, For robots Neighbors Yes The gradient.

[0030] Preferably, in S4, the distributed control strategy designed for each robot is in a fully distributed form, where each robot independently calculates the acceleration control input based on its own state and local neighbor perception information, without the need for global localization or central scheduler support.

[0031] Preferably, in S4, the final distributed control strategy expression is as follows:

[0032]

[0033] in, Represents robots The control input in the current state depends on the robot's position. and speed , is the damping coefficient.

[0034] This invention also provides a robot distributed control system based on fluid dynamics principles, and a robot distributed control method based on fluid dynamics principles applying any of the above claims, comprising:

[0035] The image acquisition module is used to acquire image information of the target area, extract the target pattern contour, construct a truncated signed range field to generate the target guiding potential energy, and calculate the target guiding force.

[0036] An initialization module is used to construct an intermediate gathering area based on the area of ​​the target pattern, and to evenly divide the robot into multiple sub-regions for initial deployment;

[0037] The interaction force construction module is used to model each robot as a particle volume element with local perception and second-order dynamic characteristics. The interaction forces between robots are constructed through smooth particle hydrodynamics, including pressure force, viscous force and density estimation.

[0038] The control strategy design module is used to design a distributed control strategy for each robot. Its control input is the sum of the target guiding force and the interaction force, and it combines damping control to stabilize the system response.

[0039] The target pattern construction module is used to control the process so that as the robot enters the target area, the target guiding force gradually decays to zero, and only the interaction force between the robots is retained to maintain the spatial uniformity of the configuration and the stability of the pattern boundary, thereby realizing the automatic construction of the target pattern.

[0040] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements a robot distributed control method based on the principle of fluid dynamics as described above.

[0041] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0042] This invention introduces a fluid dynamics modeling mechanism, combined with local sensing and short-range communication, to achieve a fully distributed pattern construction and spatial control strategy. This avoids dependence on global scheduling and spatial discretization, and possesses excellent scalability, adaptability, and control stability. The method of this invention exhibits significant advantages in spatial consistency, shape adaptability, and stability under expanded scale, making it suitable for large-scale autonomous collaborative applications across multiple platforms, including ground, air, and water. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 A flowchart illustrating the distributed control method for robots based on fluid dynamics principles provided by this invention;

[0045] Figure 2 This is a schematic diagram of the two-dimensional cubic spline kernel function used in the embodiments of the present invention;

[0046] Figure 3 These are schematic diagrams of various simulation scenarios in embodiments of the present invention;

[0047] Among them, (a) is a simulation scene of a multi-subgraph target region, (b) is a simulation scene of an octagonal star, and (c) is a simulation scene of snow.

[0048] Figure 4 This is a simulation experiment diagram of the multi-region configuration task in an embodiment of the present invention;

[0049] Figure 5 The figures show simulation results of different algorithms in a multi-subgraph scenario in this embodiment of the invention.

[0050] Figure 6 The figures show simulation results of different algorithms in different target area scenarios of different sizes in the embodiments of the present invention. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0053] like Figure 1 As shown, the present invention provides a distributed control method for robots based on the principles of fluid dynamics, comprising the following steps:

[0054] S1. Obtain image information of the target area, extract the target pattern contour, construct a truncated signed distance field to generate the target guiding potential energy, and calculate the target guiding force;

[0055] S2. Based on the area of ​​the target pattern, construct the central gathering area and evenly divide the robot into multiple sub-areas for initial deployment;

[0056] S3. Model each robot as a particle volume element with local perception and second-order dynamic characteristics, and construct the interaction forces between robots through the smooth particle fluid dynamics method, including pressure force, viscous force and density estimation.

[0057] S4. Design a distributed control strategy for each robot, with the control input being the sum of the target guiding force and the interaction force, while combining damping control to stabilize the system response.

[0058] S5. During the control process, as the robot enters the target area, the target guiding force gradually decays to zero, and only the interaction force between the robots is retained to maintain the spatial uniformity of the configuration and the stability of the pattern boundary, thereby realizing the automatic construction of the target pattern.

[0059] Specifically, this invention proposes a distributed control method for constructing complex target patterns, the overall process of which is divided into three stages:

[0060] 1. Region Segmentation and Initialization Stage: By estimating the total area of ​​the desired pattern region, an intermediate transitional clustering region is constructed using a clustering-dispersion strategy. This region is divided into several sub-regions according to the area ratio of the target pattern, achieving the initial distribution and homogenization of the robot.

[0061] 2. Guidance and Navigation Phase: During the robot migration from the cluster area to the target pattern, a potential field guiding force based on a truncated signed distance field is introduced to attract the robots to their respective target sub-regions. Simultaneously, local interaction forces based on Smooth Particle Hydrodynamics (SPH) remain constant, maintaining the compactness and stability of the formation.

[0062] 3. Local Reconstruction Stage: After the robot enters the target sub-region, the guiding force gradually weakens until it disappears. The system relies entirely on the local interaction forces such as density, pressure, viscosity and repulsion in the SPH model to maintain the spatial uniformity and boundary stability of the configuration and achieve the automatic construction of the final pattern.

[0063] Assuming the robots are arranged in a close hexagonal pattern, their sensing radius is... The unit area occupied is Define the target area The total area of ​​the target region It can be estimated as follows:

[0064]

[0065] in The total number of robots, A shape adaptive scaling factor is used to account for the geometric sparsity of the target pattern. It reflects the ratio between the effective shape area and the area of ​​its minimum enclosing rectangle.

[0066] Assuming the target region contains non-intersecting subregions Define the area of ​​the intermediate cluster region. for:

[0067]

[0068] in And satisfy .

[0069] Then gather the area Evenly divided into One corresponding partition :

[0070]

[0071] The initial distribution of each group of robots is proportional to the area of ​​the final target region.

[0072] Constructing a truncated signed distance potential function :

[0073]

[0074] In the formula, express arrive The signed Euclidean distance function of the boundary is positive outside the region and negative inside. To ensure that the potential function is zero inside the region, the positive part is truncated during actual construction, meaning that the attractive force is zero inside the region and an inward guiding force is generated outside the region.

[0075] robot Target guidance The expression is:

[0076]

[0077] in It is the target attraction gain. The target guiding force intensity factor is set to 2 by default.

[0078] To prevent robots from colliding too closely, a repulsive force is defined. :

[0079]

[0080] in, It is a repulsive gain. Indicates their relative positions. Let be the robot radius.

[0081] Smoothing kernel function With smooth length For parameters, such as Figure 2 As shown, this parameter determines the spatial support range of the kernel function. The function satisfies the normalization condition. and in Converging to Dirac The function ensures zero-order consistency. In this invention, a cubic spline kernel function in two-dimensional space is selected, and its expression is as follows:

[0082]

[0083] in, It is a normalized distance between robots. The kernel assigns higher weights to nearby robots, thereby promoting a smooth interaction field and improving numerical robustness in distributed control scenarios.

[0084] In smooth particle hydrodynamics modeling, density is typically calculated by summing the contributions of neighboring particles. However, in the scenario of swarm robot systems, the mass of individual robots is often unknown or assumed to be uniform. Therefore, we omit the mass term and instead represent the mass of each robot... density Defined as:

[0085]

[0086] Pressure corresponding to density Using the Tate state equations, it can be expressed as:

[0087]

[0088] in For reference density, This is the stiffness index, with a default value of 7. It is a pressure scaling factor related to local density. To improve computational efficiency and enhance numerical stability, it is defined as:

[0089]

[0090] in It is the pressure coefficient.

[0091] robot and viscosity terms between use Viscous form:

[0092]

[0093] in, Represents robots and The relative speed between them Indicates their relative positions. , The viscosity coefficient, Represents robots and average density, Denotes their average speed of sound, where ,item Defined as:

[0094]

[0095] Actions on robots Total interaction force based on smooth particle hydrodynamics The sum of the pressure and viscous forces exerted by all its neighboring robots:

[0096]

[0097] in For robots Neighbors Yes The gradient.

[0098] The control strategy employed in this invention is based on the physical principles of the incompressible Navier-Stokes equations, which include pressure, viscosity, and external force terms to regulate the overall evolution of the group system. In practical implementation, the original continuous physical quantities are discretized and converted into force terms determined by the local interactions between neighboring robots. A fully distributed control law is then constructed based on these force terms. The final control law combines target guiding force, interaction forces from smooth particle hydrodynamics, repulsive forces, and damping control.

[0099]

[0100] in, Represents robots The control input in the current state depends on the robot's position. and speed , This represents the damping coefficient. The physical meaning of the control input is at time... By robot The applied motion driving force will determine the robot's motion trend and behavior adjustment in the next moment. This control law can be converted into actual low-level control signals from a low-order motion model or through a speed controller, enabling continuous trajectory adjustment of the mobile robot.

[0101] To verify the adaptability and effectiveness of the control method proposed in this invention under different scenarios, a series of simulation experiments were designed. Using consistent parameter settings, the performance of three algorithms (the method of this invention, the basic artificial potential field method, and the mean drift method) in multi-region configuration tasks and the construction of target patterns of different sizes was compared. Figure 3 As shown in (a)-(c).

[0102] To systematically evaluate the coordination of group behavior and the quality of pattern construction, distance variance was selected as a performance indicator: Let... For robots The distance to its nearest neighbor, If the average of the minimum distances among all robots is taken as the mean, then the distance variance is defined as follows: The simulation parameters are shown in Table 1.

[0103] Table 1

[0104]

[0105] Figure 4 This diagram illustrates the simulation process of the method of the present invention in a multi-region configuration task. Figure 5The study further demonstrates the variation trends of distance variance for three control methods under different target patterns, which can be used to measure the spatial consistency of the robot swarm. Typical patterns such as circles, squares, triangles, stars, and rings were examined in the experiment, and the method of this invention was compared with the basic artificial potential field method and the mean drift method.

[0106] from Figure 5 As can be seen, during the entire configuration process, the method of this invention achieves significantly lower distance variance even in the aggregation stage, indicating a more compact and uniform arrangement of robots. After entering each target region, the method of this invention maintains stable and low steady-state variance under all patterns, especially in non-convex geometries, exhibiting superior boundary filling and local consistency. In contrast, comparative methods generally suffer from large fluctuations and are prone to getting trapped in local extrema, making it difficult to guarantee the overall coordination of the configuration. Experimental results show that the method of this invention outperforms existing technologies in terms of spatial consistency, shape adaptability, and convergence stability, verifying its effectiveness and robustness in complex scenarios.

[0107] Table 2

[0108]

[0109] Figure 6 The paper illustrates the spatial consistency changes of the proposed method and the comparative algorithm during the formation of two typical complex patterns (octagon and snowflake) under different group sizes (N=16~1024). The vertical axis represents the distance variance, and the horizontal axis represents the number of steps. It can be seen that as the task progresses, the distance variance of each method first increases and then decreases, reflecting the process from the initial unstructured distribution to the stable configuration of the target pattern. The proposed method exhibits faster distance variance convergence and lower steady-state variance values ​​under all test scales and pattern conditions, demonstrating superior spatial uniformity and configuration consistency. Combined with Table 2 (final group distance variance of various algorithms at different scales), it can be seen that for the octagon and snowflake target patterns, as the robot size increases, the proposed method significantly outperforms the comparative algorithm at all scales, exhibiting lower distance variance, indicating its significant advantage in maintaining local spatial consistency. Figure 6 Together with Table 2, this verifies that the method of the present invention can effectively regulate local interactions when facing large-scale complex configuration tasks, achieve stable pattern construction and distributed collaborative control, and has good system scalability and adaptability.

[0110] This invention also provides a robot distributed control system based on fluid dynamics principles, and a robot distributed control method based on fluid dynamics principles applying any of the above claims, comprising:

[0111] The image acquisition module is used to acquire image information of the target area, extract the target pattern contour, construct a truncated signed range field to generate the target guiding potential energy, and calculate the target guiding force.

[0112] The initialization module is used to construct an intermediate gathering area based on the area of ​​the target pattern and to evenly divide the robot into multiple sub-regions for initial deployment.

[0113] The interaction force construction module is used to model each robot as a particle volume element with local perception and second-order dynamic characteristics. The interaction forces between robots are constructed through smooth particle hydrodynamics, including pressure force, viscous force and density estimation.

[0114] The control strategy design module is used to design a distributed control strategy for each robot. Its control input is the sum of the target guiding force and the interaction force, and it combines damping control to stabilize the system response.

[0115] The target pattern construction module is used to control the process so that as the robot enters the target area, the target guiding force gradually decays to zero, and only the interaction force between the robots is retained to maintain the spatial uniformity of the configuration and the stability of the pattern boundary, thereby realizing the automatic construction of the target pattern.

[0116] The present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a robot distributed control method based on the principle of fluid dynamics as described above.

[0117] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0118] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for distributed control of robots based on principles of fluid dynamics, characterized in that, The method comprises the following steps: S1, acquiring image information of a target region, extracting a contour of a target pattern, and constructing a truncated signed distance field to generate a target guide potential energy, and calculating a target guide force; S2, constructing an intermediate aggregation region according to an area of the target pattern, and uniformly dividing the robots into a plurality of sub-regions for initialization and deployment; S3, modeling each robot as a particle body element with local perception and second-order dynamic characteristics, constructing interaction forces between the robots by a smoothed particle hydrodynamics method, including pressure forces, viscous forces and density estimation; S4, designing a distributed control strategy for each robot, the control input of which is the sum of the target guide force and the interaction force, and combining damping control to stabilize the system response; S5, in the control process, as the robots enter the target region, the target guide force gradually decays to zero, and only the interaction force between the robots is retained to maintain the spatial uniformity of the configuration and the stability of the pattern boundary, thereby realizing automatic construction of the target pattern.

2. The robot distributed control method based on fluid dynamics principles according to claim 1, characterized in that, The In some embodiments, the target guiding force is calculated from the gradient direction of the truncated signed distance field corresponding to the target pattern, and has the ability to attract the robot into the target region. The target guiding force is calculated according to the following formula: wherein, is a target attraction gain, is a target guidance force strength factor, is a target guidance force, denotes a non-negative distance field to the target region boundary.

3. The robot distributed control method based on fluid dynamics principles according to claim 1, characterized in that, The In the robot interaction force is weighted by a two-dimensional cubic spline kernel function, the two-dimensional cubic spline kernel function Satisfies the normalization and compactness: where, is the normalized inter-robot distance, is the smoothing length; the kernel gives higher weights to nearby robots.

4. The robot distributed control method based on fluid dynamics principles according to claim 1, wherein, The In the embodiment, the interaction force between the robots is constructed by a smoothed particle hydrodynamics method, specifically including: Robots Density of the particles is defined as: wherein, is a kernel function, is a robot position, is a robot position; density corresponds to pressure is expressed using the Tait equation of state as: wherein, is the reference density, is the stiffness index, is a pressure scaling factor related to the local density; robot and viscous term between adopting viscous form: where denotes the relative velocity between the robot and , denotes the relative position between the robot and , , is the viscosity coefficient, denotes the average density of the robot and , denotes their average sound speed, where , the term is an approximation of the velocity divergence term, defined as: ; Acting on robots Total interaction forces based on smoothed particle hydrodynamics The sum of pressure forces and viscous forces generated for all neighboring robots wherein, is a neighbor of the robot is a gradient of is a gradient of is a gradient of 5. The robot distributed control method based on fluid dynamics principles according to claim 1, wherein, The In this paper, a fully distributed control strategy is designed for each robot, which computes the acceleration control input independently according to its own state and local neighbor perception information without global localization or central dispatcher support.

6. The robot distributed control method based on fluid dynamics principles according to claim 5, wherein, The In the final configuration, the distributed control strategy expression is as follows: wherein representing a robot the control input in the current state, depends on the position and velocity of the robot is a damping coefficient.

7. A robot distributed control system based on fluid dynamics principles, applying a robot distributed control method based on fluid dynamics principles according to any one of claims 1 to 6, characterized in that, The method comprises: An image acquisition module for acquiring image information of a target region, extracting a contour of a target pattern, and constructing a truncated signed distance field to generate a target guide potential energy, and calculating a target guide force; An initialization module for constructing an intermediate aggregation region according to an area of the target pattern, and uniformly dividing the robots into a plurality of sub-regions for initialization and deployment; An interaction force construction module for modeling each robot as a particle body element with local perception and second-order dynamic characteristics, and constructing interaction forces between the robots by a smoothed particle hydrodynamics method, including pressure forces, viscous forces and density estimation; A control strategy design module for designing a distributed control strategy for each robot, the control input of which is the sum of the target guide force and the interaction force, and combining damping control to stabilize the system response; A target pattern construction module for, in the control process, as the robots enter the target region, the target guide force gradually decays to zero, and only the interaction force between the robots is retained to maintain the spatial uniformity of the configuration and the stability of the pattern boundary, thereby realizing automatic construction of the target pattern.

8. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by a processor to implement a distributed robot control method based on fluid dynamics principles according to any one of claims 1 to 6.

Images