A method for generating multiple patterns of self-organizing swarm robots

Abstract

This invention provides a method for generating multiple patterns in a self-organizing swarm robot group, belonging to the field of swarm robot pattern formation control. The method includes selecting the number of robots participating in pattern generation and distributing the robots sequentially or randomly within a certain area of a scene; assigning a state transition cycle to each robot; and assigning the same position update rule to each robot based on the state transition cycle. Within the sensing range, robots react behaviorally based on the states and positions of neighboring robots, thereby completing the generation of multiple patterns in the swarm robot group. This invention addresses the problems of existing swarm robot pattern generation technologies, such as high computational and communication costs, a single pattern, poor robustness and scalability, and the use of a global coordinate system and global information.

Description

Technical Field

[0001] The present invention belongs to the field of group robot pattern formation control, and in particular relates to a multi-pattern generation method for self-organizing group robots. Background Art

[0002] Swarm robots consist of large numbers (often hundreds or thousands) of simple autonomous robots with limited perception and communication capabilities. Individually, they are incapable of performing any task independently. Instead, they leverage local interactions between individuals to stimulate collective behavior and accomplish large-scale, complex tasks, such as regional exploration, coordinated transport, spatial deployment, and formations. Compared to highly intelligent individual robots, swarm robots offer flexibility, scalability, and robustness, and therefore have attracted widespread attention. Since 2014, numerous countries have launched drone swarm research projects, such as LOCUST, OFFSET, GREMLINS, and CODE, focusing on key technical challenges such as distributed cluster formations, coordinated flight, and target encirclement. However, swarm robots typically organize themselves in either a tightly or loosely organized manner, relying on a specific overall form to accomplish specific tasks. Therefore, pattern generation methods for swarm robots are a core issue in swarm robotics research.

[0003] A variety of swarm robotic pattern formation methods have been proposed. From the perspective of macroscopic and microscopic inspiration from swarm organisms, existing swarm robotic pattern formation methods can be divided into those based on multicellular differentiation and those based on the macroscopic behavior of biological swarms. Among these multicellular differentiation-based pattern formation methods are morphological diffusion, reaction-diffusion, and gene regulatory networks. These methods achieve distributed pattern formation in swarm robots and exhibit a certain degree of robustness and scalability. These methods even do not require mutual position information. However, their core challenges remain unresolved. Multicellular differentiation-based pattern formation methods suffer from poor reliability, accuracy, and slow convergence. Even with the same parameters, the resulting patterns can vary significantly, as demonstrated in the work of Cotterell and Slavkov. Furthermore, while these methods can achieve swarm robotic pattern generation, the resulting patterns are simple and limited in variety. Methods based on the macroscopic behavior of biological populations can be divided into structural methods and behavioral methods. Algorithms based on such methods are simple to construct, the pattern patterns they form are fast and accurate, and the communication costs are low. However, the core problem is their poor robustness, and more work relies on global information or a global coordinate system. In addition, some algorithms still require direct or indirect pre-setting of pattern patterns, such as the work of Michael, Xu, MacDonald, Tan, and Cheah. Summary of the Invention

[0004] In response to the above-mentioned deficiencies in the prior art, the present invention provides a method for generating multiple patterns of self-organizing swarm robots, which has the advantages of simple control method, low communication cost, fast pattern convergence speed, stable pattern, good repeatability and certain robustness, and does not require global information and global coordinate system.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0006] This solution provides a method for generating multiple patterns of self-organizing swarm robots, comprising the following steps:

[0007] S1. Select the number of robots to participate in pattern generation and distribute the robots in a certain area of the scene in sequence or randomly;

[0008] S2, assigning a state change cycle to each robot;

[0009] S3, based on the state change cycle, assign the same position update rule to each robot;

[0010] S4. Within the sensing range, the robot responds with behavior based on the status and position of the neighboring robots to complete the generation of multiple patterns of the swarm robots.

[0011] The beneficial effects of the present invention are as follows: the present invention provides a multi-pattern generation method for swarm robots, wherein the robots have the characteristics of state switching, wherein different pattern generation can be achieved by changing the state transformation cycle, and robots within the perception range make different reaction behaviors according to the state and position of neighboring robots, and the robot state can be recognized by other robots through direct or indirect communication, thereby reducing the communication cost, and does not require global information and a global coordinate system, and has great practicality in actual application; the robot position update rule described in the present invention is simple, and the calculation cost of the robot is greatly reduced; in addition, the robot position update rule only depends on the state and position of neighboring robots in a local range, and has strong scalability.

[0012] Furthermore, the step S2 is specifically as follows:

[0013] Each robot is assigned a state change cycle. The robot state is switched every other state change cycle. The state is represented by a state value, and one state corresponds to a unique state value.

[0014] The beneficial effects of the above further scheme are: the robot state can be easily realized through simple physical technology or built-in programs. The change of state gives the robot the ability to drive itself, that is, to take corresponding actions by sensing the state of its neighbors. The behavior can be changed according to the needs of the task, so it can be used as an extensible behavior framework.

[0015] Furthermore, the expression of the position update rule in step S3 is as follows:

[0016]

[0017]

[0018]

[0019] Among them, F i-t It represents the net force on robot i at time step t, which is a vector, n i represents the total number of neighbor robots within the sensing or communication range, q represents a non-zero constant coefficient, z i and z j They represent the state values of robot i and robot j respectively. a, b and c represent the virtual force adjustment constants, which are used to adjust the size of the robot's attraction and repulsion areas. ij represents the direction vector from robot j to robot i, l ij represents the distance between robot i and robot j, T i and T j Represent the state change cycles of robot i and robot j respectively.

[0020] The beneficial effect of this further solution is that robots can attract or repel each other by sensing the distance and state of their neighbors. This computation is simple and fast, does not consume excessive computing memory, and avoids the use of global information. Depending on the state representation, communication costs can be reduced or even eliminated. The use of relative positions in this solution avoids the use of a global coordinate system, making the physical implementation of swarm robots simpler.

[0021] Furthermore, the movement of the robot follows the cellular automaton model, and the cell closest to the robot in the direction of the resultant force acting on the robot at time step t is used as the updated position at time step t+1. If the updated position of the robot at time step t+1 is occupied by other robots, the robot waits until the next time step.

[0022] The beneficial effects of the above further solution are: the present invention is based on a cellular automaton model, simplifies the kinematic model, and can effectively avoid collisions between robots by waiting until the next time step. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Flow chart of the method of the present invention.

[0024] Figure 2 Schematic diagram of the working scene of the swarm robot in this embodiment.

[0025] Figure 3Schematic diagram of the state switching cycle of the swarm robot in this embodiment.

[0026] Figure 4 Schematic diagram of robot state switching in this embodiment.

[0027] Figure 5 This is a schematic diagram of the behavior of the robot in this embodiment towards robots in different states within its perception range.

[0028] Figure 6 Schematic diagram of the swarm robot motion model based on cellular automata in this embodiment.

[0029] Figure 7 Schematic diagram of simulation results in this embodiment.

[0030] Figure 8 This is a result diagram of the reliability and stability verification experiment provided in this embodiment.

[0031] Figure 9 This is a result diagram of the second reliability and stability verification experiment provided in this embodiment.

[0032] Figure 10 The third result of the reliability and stability verification experiment provided in this embodiment is shown in FIG.

[0033] Figure 11 This is a diagram of the robustness verification experiment results provided in this embodiment.

[0034] Figure 12 This is a result diagram of the multi-pattern generation capability verification experiment mentioned in this embodiment.

[0035] Figure 13 This is the result of the second experiment to verify the multi-pattern generation capability proposed in this embodiment.

[0036] Figure 14 This is the result of the third experiment to verify the multi-pattern generation capability proposed in this embodiment. DETAILED DESCRIPTION

[0037] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0038] Example

[0039] like Figure 1 As shown, the present invention provides a method for generating multiple patterns of self-organizing group robots, and its implementation method is as follows:

[0040] S1. Select the number of robots to participate in pattern generation and distribute the robots in a certain area of the scene in sequence or randomly;

[0041] In this embodiment, Figure 2 As shown in the figure, 500 robots are selected and arranged in sequence in a certain area of the scene.

[0042] S2. Assign a state change cycle to each robot, which is as follows:

[0043] Each robot is assigned a state change cycle. The robot state is switched every other state change cycle. The state is represented by a state value, and one state corresponds to a unique state value.

[0044] In this embodiment, Figure 3 As shown in , each robot is given a state change cycle. The initial state of the robots is the same, and the robot state switches once every state change cycle, as shown in Figure 3 Preferably, the state is represented by a state value of -1 or 1.

[0045] In this embodiment, the robots may be divided into groups with priority, and a subgroup adopts the same state change cycle.

[0046] In this embodiment, the robot has two states: The state can be represented physically, such as the color of the robot's built-in light, or directly by a digital signal stored in the robot's memory; the state is represented by a state value, with each state corresponding to a unique state value. The robot has a limited perception and communication range and can identify the location of neighboring robots within its perception range. The robot can transmit its state to neighboring robots via physical or digital signals, which the neighboring robots recognize and ultimately convert into a state value. The robot can switch between the two states, changing state every certain time step.

[0047] S3. Based on the state change cycle, assign the same position update rule to each robot:

[0048]

[0049]

[0050]

[0051] Among them, F i-t It represents the net force on robot i at time step t, which is a vector, n i represents the total number of neighbor robots within the sensing or communication range, q represents a non-zero constant coefficient, zi and z j They represent the state values of robot i and robot j respectively. a, b and c represent the virtual force adjustment constants, which are used to adjust the size of the robot's attraction and repulsion areas. ij represents the direction vector from robot j to robot i, l ij represents the distance between robot i and robot j, T i and T j Represent the state change cycles of robot i and robot j respectively.

[0052] S4. Within the sensing range, the robot responds behaviorally based on the state and position of the neighboring robots to complete the generation of the swarm robot multi-pattern, wherein the cell closest to the robot in the direction of the resultant force at time step t is used as the updated position at time step t+1. If the updated position of the robot at time step t+1 is occupied by other robots, the robot waits until the next time step.

[0053] In this embodiment, the movement of the robot follows the cellular automaton model. At time step t, the robot occupies a cellular lattice. The robot can select the adjacent cell of the current cellular lattice as the update position at time step t+1.

[0054] In this embodiment, when robot i calculates the net force acting on it, it selects the nearest cell in the direction of the net force as the updated position for time step t+1 in the next time step. At the current time step, if robot i recognizes that the updated position is already occupied by another robot, it waits until the next time step.

[0055] In this embodiment, the motion rules of the robot i described above are also applicable to other robots used. The robots in the cluster are homogeneous, and their capabilities and execution programs are completely consistent.

[0056] In this embodiment, Figure 5 As shown, the robot responds behaviorally to robots in different states and positions within its perception range, and the behavioral responses are characterized by attraction and repulsion.

[0057] In this embodiment, the movement of the robot follows the cellular automaton model. When a robot occupies a cell, the current cell has 8 adjacent cells (C1-C8). Figure 6 When the robot's updated position is already occupied by another robot, it waits until the next time step. According to the position update rule, the robot will calculate the resultant force on itself at time step t, and the adjacent cell pointed by the resultant force will be used as the updated position at time t+1.

[0058] In this embodiment, before the simulation results are presented, some assumptions are explained in the simulation: the robots are simplified to point masses, and their motion is regarded as point mass motion; the robots have a limited perception range, and a local coordinate system is used to describe the positional relationship between the robots; and the robots are homogeneous.

[0059] According to the above conditions and rules, the pattern formation process of 500 robots was obtained based on Matlab simulation software, as shown in Figure 7 As shown. Figure 2 As the time steps progress, the robot states continuously switch, allowing the entire group to adaptively make corresponding dynamic adjustments. Finally, at t = 276, the pattern stabilizes and the entire group no longer undergoes significant changes, and the pattern converges. In addition, some meaningful patterns also appear during the pattern convergence process. The transition of the robot state breaks the internal dynamic balance of the original pattern, and the pattern switches from one form to another, such as Figure 7 (a)-(c), The intrinsic motivation of pattern morphology switching is closely related to the robot's state transition.

[0060] In this example, simulation experiments were conducted to verify the reliability and stability of the proposed method. Specifically, the authors verified whether the population patterns obtained from repeated experiments under the same parameters produced similar results and converged to a stable pattern. The similarity of the patterns was characterized using fractal dimension, a technique used in fractal theory. The fractal dimension was calculated using the commonly used box counting method. The simulation program was consistent with the above-described example.

[0061] In this embodiment, the simulation results are as follows: Figures 8-10 As shown, the same simulation program was repeated three times. From a sensory perspective, the resulting patterns formed by the swarm robots are almost identical. The fractal dimension calculation results show that the corresponding fractal dimensions during the pattern formation process are relatively small, as shown in Table 1. The fractal dimensions of the patterns at the final convergence are very similar, thus verifying the good reliability and stability of the proposed method.

[0062] Table 1

[0063]

[0064] In this embodiment, the robustness of the method proposed in the present invention is verified through simulation experiments. Taking 500 robots as an example, during the pattern formation process, at specific moments, the robots are randomly destroyed and made invalid, and the pattern is observed to see whether it can still maintain its original shape. During the simulation process, the robots are randomly destroyed and made invalid at time step t=200, and the failure rates are 5%, 10%, 15%, and 30% of the total number of robots respectively. The simulation results are shown in Figure 2. Figure 11As shown in the figure, the results show that when the robot failure rate is 30%, the pattern of the swarm robot does not change much, which verifies that the multi-pattern generation method of the self-organizing swarm robot proposed in the present invention has good robustness.

[0065] In this embodiment, the robot state change cycle is modified through simulation experiments to verify that the method proposed in the present invention has the ability to generate multiple patterns. Figure 12-14 The patterns corresponding to different pass transformation periods are displayed, thereby verifying that the multi-pattern generation method of a self-organizing swarm robot proposed in the present invention has the ability to generate multiple patterns.

[0066] In summary, the present invention proposes a method for generating multiple patterns for self-organizing swarm robots. This method is based on the behavioral approach and combines the cellular automaton model to construct local rules for the robots. The reliability, stability, and robustness of the swarm robots are evaluated through simulation analysis. In addition, the swarm robot pattern generation framework proposed in the present invention avoids the use of global information and global coordinates and does not require predefined patterns. The framework proposed in the present invention selects a communication method based on actual conditions and requires a small communication capacity, thereby reducing communication costs. In addition, the method proposed in the present invention is computationally simple and does not require high computing power from the robots, indirectly improving the convergence speed of the entire swarm.

[0067] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A method for generating multiple patterns of self-organizing swarm robots, characterized in that: The following steps are involved: S1. Select the number of robots to participate in pattern generation and distribute the robots in a certain area of the scene in sequence or randomly; S2, assigning a state change cycle to each robot; S3, based on the state change cycle, assign the same position update rule to each robot; The expression of the position update rule is as follows: in, Represents a robot At time step The net force is a vector, represents the total number of neighbor robots within the sensing or communication range, represents a constant coefficient that is not 0, and Represents robots and robots The status value of 、 and Represents the virtual force adjustment constant, which is used to adjust the size of the robot's attraction and repulsion area. Indicates that from the robot Pointing robot The direction vector, Represents a robot and robots The distance between and Represents robots and robots The state change cycle; S4. Within the sensing range, the robot responds with behavior based on the status and position of the neighboring robots to complete the generation of multiple patterns of the swarm robots.

2. The method for generating multiple patterns of self-organizing swarm robots according to claim 1, characterized in that: The step S2 is specifically as follows: Each robot is assigned a state change cycle. The robot state is switched every other state change cycle. The state is represented by a state value, and one state corresponds to a unique state value.

3. The method for generating multiple patterns of self-organizing swarm robots according to claim 1, characterized in that: The robot's motion follows a cellular automaton model, and the robot moves at time steps. The nearest cell in the direction of the resultant force is taken as the time step The updated position of the robot is If the updated position is occupied by other robots, wait until the next time step.

Images