A method, apparatus, electronic device and storage medium for obstacle avoidance in a multi-agent swarm

Abstract

This application provides a multi-agent swarm obstacle avoidance method, apparatus, electronic device, and storage medium. The method includes: acquiring initial dynamic parameters of the agent swarm comprising multiple agents, the initial dynamic parameters including displacement, velocity, and initial control input; determining multiple second agents within a preset range of a first agent based on the initial dynamic parameters, the first agent being any one of the multiple agents; acquiring motion parameters of the first agent and observed obstacle positions; determining a first control input for the first agent based on the motion parameters, obstacle positions, and initial dynamic parameters; and performing obstacle avoidance simulation verification on the first agent based on the first control input. This application relates to the technical field of swarm control and can improve the practicality of multi-agent swarm obstacle avoidance algorithms.

Description

Technical Field

[0001] This application relates to the technical field of cluster control, specifically to a method, apparatus, electronic device, and storage medium for obstacle avoidance in a multi-agent cluster. Background Technology

[0002] An unmanned system swarm is a collaborative system composed of multiple autonomous or semi-autonomous intelligent agents, encompassing various forms of unmanned equipment such as drones, unmanned surface vessels, unmanned underwater vehicles, unmanned vehicles, and mobile robots. These intelligent agents interact and collaborate through networks to jointly accomplish complex tasks. This type of system is also known as a "multi-agent swarm," and its key characteristic is that each intelligent agent in the swarm possesses a certain degree of independent decision-making ability, enabling them to collaborate, communicate, and make decisions in a distributed manner within the group, thereby optimizing overall goals and improving efficiency.

[0003] Obstacle avoidance is one of the core challenges in multi-agent swarm motion control, especially in complex environments. In 1986, Reynolds proposed three basic principles for multi-agent swarm motion: aggregation, collision avoidance, and velocity matching. The aggregation principle requires each agent to move as close as possible to its neighbors to maintain group cohesion; the collision avoidance principle ensures that agents maintain sufficient distance to avoid collisions; and the velocity matching principle emphasizes the consistency of velocity among neighboring agents to maintain swarm coordination. When the swarm encounters obstacles during movement, in addition to following obstacle avoidance strategies to avoid collisions, it must also continue to satisfy these three basic principles throughout the overall obstacle avoidance process.

[0004] Traditional flopping algorithms are widely used due to their robustness in swarm control, particularly excelling in ensuring swarm member aggregation, preventing collisions, and maintaining speed consistency. However, flopping algorithms exhibit limitations when facing obstacle avoidance requirements in dynamic and complex environments. To achieve obstacle avoidance, artificial potential field methods are often used. This method relies on pre-modeling of the global environment, calculating virtual force fields between agents and obstacles to guide them in avoiding obstacles. Artificial potential field methods are highly sensitive to environmental changes; when the environment changes, the algorithm and model often need to be redesigned, resulting in poor adaptability and difficulty in handling dynamic or unknown environments. Therefore, a method is needed to improve the practicality of multi-agent swarm obstacle avoidance algorithms. Summary of the Invention

[0005] This application provides a multi-agent cluster obstacle avoidance method, apparatus, electronic device, and storage medium, which can improve the practicality of multi-agent cluster obstacle avoidance algorithms.

[0006] The first aspect of this application provides a multi-agent swarm obstacle avoidance method, the method comprising:

[0007] The initial dynamic parameters of the intelligent agent cluster, which comprises multiple intelligent agents, are obtained. These initial dynamic parameters include displacement, velocity, and initial control input.

[0008] Based on the initial dynamic parameters, a plurality of second agents within a preset range of the first agent are determined, wherein the first agent is any one of the plurality of agents.

[0009] Obtain the motion parameters of the first intelligent agent and the observed positions of obstacles;

[0010] Based on the motion parameters, the obstacle position, and the initial dynamic parameters, the first control input of the first intelligent agent is determined;

[0011] Based on the first control input, obstacle avoidance simulation verification is performed on the first intelligent agent.

[0012] Optionally, determining the first control input of the first intelligent agent based on the motion parameters, the obstacle position, and the initial dynamic parameters specifically includes:

[0013] Based on the first velocity of the first agent, the first displacement of the first agent, the second velocity of the second agent, and the second displacement of the second agent, the collision avoidance control input of the first agent is determined, and the collision avoidance control input is expressed as follows:

[0014]

[0015] Among them, f i g q is the collision avoidance control input. i p represents the first displacement of the first intelligent agent. i Let q be the first velocity of the first intelligent agent. j p represents the second displacement of the j-th second agent. j Let c1 and c2 be the second velocity of the j-th second agent, and c1 and c2 be the control parameters of the first agent. ij Let a be the displacement vector from the displacement of the first agent to the displacement of the j-th second agent. ij Φ is the interaction strength coefficient. α It is a distance function;

[0016] Obtain the target point position and target point velocity of the intelligent agent cluster;

[0017] Based on the target point position and the target point velocity, the initial control input is determined as follows:

[0018] f in =-e1(q i -q target )-e2(p i -p target )

[0019] Among them, f i n q is the initial control input. i For the first displacement, p i Let q be the first velocity. target p is the location of the target point. target The target point velocity is given by e1 and e2, which are control parameters.

[0020] Based on the collision avoidance control input and the initial control input, the first control input is determined, and the first control input is represented as follows:

[0021]

[0022] Among them, u i f is the first control input. i g To avoid collisions between multiple intelligent agents, f is the collision avoidance control input. i n f is the initial control input. i β To control obstacle avoidance input.

[0023] Optionally, determining the first control input based on the obstacle avoidance control input further includes:

[0024] Based on the first displacement and the first velocity, the observation space is determined as follows:

[0025]

[0026] Among them, s i For observation space, The first displacement and the first velocity are, S represents the observation of the first agent on the second agent. i obstacle This refers to the first intelligent agent's observation of obstacles;

[0027] Based on the observation space, the control obstacle avoidance input is determined, and the control obstacle avoidance input is represented as follows:

[0028]

[0029] Among them, f iβ Δp is the control input for obstacle avoidance. i The change in velocity is the amount of velocity change.

[0030] Optionally, determining the collision avoidance control input of the first intelligent agent based on the first velocity, the first displacement of the first intelligent agent, the second velocity of the second intelligent agent, and the second displacement of the second intelligent agent further includes:

[0031] Based on the first displacement and the second displacement, the displacement vector is determined, and the displacement vector is specifically represented as follows:

[0032]

[0033] Where, n ij Let q be the displacement vector. i Let q be the first displacement of the first intelligent agent. j This represents the second displacement of the j-th second agent;

[0034] Based on the first displacement and the second displacement, the interaction strength coefficient is determined, specifically calculated using the following formula:

[0035]

[0036] Among them, a ij Let q be the interaction strength coefficient. j -q i || σ It means (q) j -q i The σ norm of q i Let q be the first displacement of the first intelligent agent. j For the second displacement of the j-th second agent, the σ norm is defined as follows:

[0037]

[0038] Where, the constant ε>0, ‖x‖ σ Let σ represent the σ norm of vector x, and ‖x‖ represent the length of the vector, where x is a vector of arbitrary dimensions.

[0039] Optionally, after determining the first control input of the first agent based on the motion parameters, the obstacle position, and the initial dynamic parameters, the method further includes:

[0040] Acquire multiple obstacles observed by the first intelligent agent and multiple second intelligent agents;

[0041] Determine the distance between each of the obstacles and the first intelligent agent;

[0042] Identify a target obstacle among multiple obstacles, the target obstacle being closest to the first agent;

[0043] A reward function is established based on the distance between the target obstacle and the first agent, and the reward function is as follows:

[0044]

[0045] Where, r i Let d be the reward function. i d1 is the distance between the first agent and the target obstacle, μ is the repulsion coefficient, and d0 is the reward adjustment scale.

[0046] Optionally, the distance function is determined based on the distance between the first agent and the second agent, specifically expressed by the following formula:

[0047]

[0048] Where, Φ α (z) is a distance function with distance z as the independent variable, z = ||q j -q i ||,r a For the normalization scale, d a This is a constant offset.

[0049] Optionally, the step of performing obstacle avoidance simulation verification on the first intelligent agent based on the first control input specifically includes:

[0050] Construct a simulation environment, which includes at least two virtual obstacles and virtual target points;

[0051] A virtual intelligent agent model is constructed based on the first control input;

[0052] The virtual intelligent agent interacts within the simulation environment;

[0053] Determine whether the virtual agent can avoid the virtual obstacles and reach the virtual target point;

[0054] If it is determined that the virtual agent avoids the virtual obstacle and reaches the virtual target point, then the first control input is determined to be an obstacle avoidance strategy.

[0055] A second aspect of this application provides a multi-agent swarm obstacle avoidance device, the device comprising an acquisition module, a judgment module, a processing module, and a simulation module, wherein:

[0056] The acquisition module is used to acquire the initial dynamic parameters of the intelligent agent cluster, which includes multiple intelligent agents. The initial dynamic parameters include displacement, velocity, and initial control input.

[0057] The judgment module is used to determine a plurality of second agents within a preset range of the first agent based on the initial dynamic parameters, wherein the first agent is any one of the plurality of agents;

[0058] The acquisition module is used to acquire the motion parameters of the first intelligent agent and the observed obstacle positions;

[0059] The processing module is used to determine the first control input of the first intelligent agent based on the motion parameters, the obstacle position, and the initial dynamic parameters;

[0060] The simulation module is used to perform obstacle avoidance simulation verification on the first intelligent agent based on the first control input.

[0061] A third aspect of this application provides an electronic device including a processor, a memory, a user interface, and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any of the foregoing.

[0062] A fourth aspect of this application provides a computer-readable storage medium storing instructions that, when executed, perform the method described in any of the preceding descriptions.

[0063] In summary, one or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0064] 1. This application dynamically acquires and analyzes the initial dynamic parameters of the agent, the positions of other agents and obstacles within a preset range, and combines this with the agent's motion parameters to dynamically determine the agent's control input, effectively addressing complex and dynamic environments. By verifying the obstacle avoidance effect of this control input in a simulation environment, it avoids static presets and dependencies on the global environment, significantly improving the algorithm's adaptability and flexibility. Furthermore, by adaptively adjusting the agent's behavior, it reduces dependence on the environmental model, thereby improving the obstacle avoidance performance and practicality of multi-agent clusters in uncertain and dynamic environments.

[0065] 2. The obstacle avoidance method proposed in this invention is applicable to situations where the environment is not fully known, avoiding the need for modeling the global environment. By having the agent observe its local environment and restrict the observation space to a small range, this method facilitates rapid learning and training, improving efficiency. Simultaneously, this method uses the agent's velocity changes as the action space for reinforcement learning, aligning with practical control requirements. During the interaction between the agent and the local environment, reinforcement learning strategies enable the agent to gradually explore and effectively avoid obstacles, thereby enhancing the system's applicability and scalability.

[0066] 3. This application enhances the obstacle avoidance capability of multi-agent clusters in unknown and complex environments by dynamically generating control inputs based on the agent's local observation space, motion parameters, and obstacle positions. It avoids the complexity and uncertainty of global environment modeling, enabling each agent to make real-time decisions based on local information. By combining collision avoidance control, initial control input, and obstacle avoidance control input based on the observation space, this method effectively improves the agent's reaction speed and adaptability in dynamic environments. Furthermore, the design of reward and distance functions further optimizes the agent's obstacle avoidance behavior, ensuring stability and safety in complex environments, and demonstrating strong practicality and scalability. Attached Figure Description

[0067] Figure 1 This is a flowchart illustrating a multi-agent cluster obstacle avoidance method disclosed in an embodiment of this application;

[0068] Figure 2 This is a schematic diagram of the initial simulation process of a multi-agent cluster obstacle avoidance method disclosed in an embodiment of this application;

[0069] Figure 3 This is a schematic diagram of the simulated obstacle avoidance process of a multi-agent cluster obstacle avoidance method disclosed in an embodiment of this application;

[0070] Figure 4 This is a schematic diagram illustrating the simulation results of a multi-agent cluster obstacle avoidance method disclosed in an embodiment of this application.

[0071] Figure 5 This is a schematic diagram of a multi-agent cluster obstacle avoidance device disclosed in an embodiment of this application;

[0072] Figure 6 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application.

[0073] Explanation of reference numerals in the attached figures: 501, Acquisition module; 502, Judgment module; 503, Processing module; 504, Simulation module; 601, Processor; 602, Communication bus; 603, User interface; 604, Network interface; 605, Memory. Detailed Implementation

[0074] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0075] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0076] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0077] Unmanned system swarms consist of multiple autonomous or semi-autonomous agents that collaborate through a network to complete complex tasks and possess independent decision-making capabilities. Obstacle avoidance is one of the core challenges in multi-agent swarm control. Reynolds' three principles—aggregation, collision avoidance, and velocity matching—provide a foundation for swarm motion control. However, while traditional flopping algorithms are robust in maintaining swarm aggregation and velocity consistency, they exhibit limitations in obstacle avoidance in dynamic and complex environments. They typically rely on artificial potential fields for global environment modeling, resulting in poor adaptability. Therefore, it is necessary to combine new methods such as reinforcement learning to enhance the flexibility and practicality of swarm obstacle avoidance algorithms.

[0078] This embodiment discloses a multi-agent swarm obstacle avoidance method, referring to... Figure 1 It includes the following steps:

[0079] S110: Obtain the initial dynamic parameters of the agent cluster, which contains multiple agents.

[0080] Dynamic modeling of a multi-agent swarm is fundamental to the entire system, describing the motion state of each agent and its changing patterns. Assuming a swarm of N agents, each agent has its own motion state (displacement, velocity) and control inputs. Agents can interact through perception or communication, and each agent has a certain degree of awareness of its surrounding environment (including other agents and obstacles).

[0081] For the i-th agent, its motion state is described by the following two dynamic equations:

[0082]

[0083] Where, q i p represents displacement. i u represents velocity. i This represents the control input of the i-th agent.

[0084] The agent's displacement q i Its speed p i The integral of . In discrete time, the equation can be expressed as:

[0085] q i (t+1)=q i (t)+p i (t)·Δt

[0086] Where Δt is the time step. Similarly, the agent's velocity p i Its control input u i The decision. Similarly, in discrete time, it can be expressed as:

[0087] p i (t+1)=p i (t)+u i (t)·Δt

[0088] The state of each agent consists of its displacement qi and velocity pi, therefore the state of the i-th agent can be represented as x. i =[q i p i ] T The state space of the entire multi-agent system is a high-dimensional space composed of the states of all agents. The system's state vector can be represented as X = [x1, x2, ..., x...]. N ] T .

[0089] The control input for each agent is determined by a variety of factors, including the states of other agents, the target position, and obstacle avoidance requirements. Control input design can utilize traditional control methods (such as PID control) or combine them with artificial intelligence methods (such as reinforcement learning) to design more intelligent control laws. In multi-agent systems, a distributed control strategy is typically employed, where each agent independently calculates its own control input without relying on global information. This strategy not only improves system robustness but also better adapts to dynamically changing environments.

[0090] S120, based on the initial dynamic parameters, determine multiple second agents within a preset range of the first agent.

[0091] In the initial state of a multi-agent cluster, the dynamic parameters (including displacement, velocity, control input, etc.) of each agent need to be acquired through sensing or communication. For example, for any one agent, the initial dynamic parameters of the first agent include its position, velocity, and control input.

[0092] Each agent obtains its current position through GPS or other positioning systems, its current velocity through an inertial measurement unit (IMU) or other velocity measurement devices, shares information with other agents through wireless communication networks, and obtains the state data of neighboring agents.

[0093] An intelligent agent uses an observation space to identify multiple neighboring agents within a predefined range. The observation space is the set of environmental information that the agent can perceive or acquire. This space typically includes its own state information and observation information about other agents and obstacles. Here, the observation space mainly comprises the following three parts: its own state information, including the agent's displacement and velocity; observation information about other agents, including the positions and velocities of neighboring agents; and observation information about obstacles, including their positions.

[0094] In this embodiment, it is assumed that the observation range of each agent is large enough, but to simplify processing, each agent only observes its three neighboring agents. The process for selecting neighboring agents is as follows:

[0095] For the first agent, we first need to calculate its distance from other agents. Let the position of the j-th agent be q. j Then the Euclidean distance d between the first agent and the j-th agent is... ij for:

[0096] d ij =||q i -q j ||

[0097] Based on the calculated distance dij, select the multiple agents closest to the i-th agent to form the neighbor agent set of that agent. Sort all agents in ascending order of distance to obtain the nearest neighbor agent list. Select the multiple closest agents from the sorted results to obtain multiple second agents. The second agents are the agents within a preset range of the first agent among the multiple agents.

[0098] Although it is assumed that the agent's observation range is large enough, in practical applications, the agent's observation range is limited due to sensor and communication limitations. If an agent has more than a predetermined number of neighboring agents (e.g., 3), the 3 closest agents can be selected first. If there are fewer than 3 neighboring agents, only the agents that are actually observed are selected.

[0099] S130, acquire the motion parameters of the first agent and the observed positions of obstacles.

[0100] To avoid collisions, an intelligent agent needs to perceive the location of obstacles in its surrounding environment. Obstacle observation can be achieved in the following ways: the agent's lidar emits laser beams and receives reflected signals to generate distance and orientation information of the obstacles, typically used to create high-precision two-dimensional or three-dimensional environmental maps. This is currently a widely used obstacle detection sensor. Alternatively, the agent's ultrasonic sensors emit and receive ultrasonic signals to measure the distance between the agent and the obstacle, often used in low-cost, low-complexity obstacle avoidance systems.

[0101] After an obstacle is observed, its position needs to be estimated. Common obstacle position estimation methods include: dividing the environment into a grid, where each grid represents whether a certain area is occupied by an obstacle. The agent updates the grid state based on sensor data, thereby estimating the obstacle's position. Vision-based object detection algorithms (such as YOLO and SSD) are used to identify obstacles in the image, and methods such as Kalman filtering are used to track the obstacle's trajectory, thus continuously estimating its position. In dynamic environments, obstacle positions may change over time. The agent needs to continuously observe and update obstacle position information in real time. This can be achieved by periodically scanning the environment and updating the obstacle map. For dynamic obstacles (such as moving pedestrians or vehicles), the agent also needs to predict their future positions to plan safe obstacle avoidance paths.

[0102] S140, based on motion parameters, obstacle positions, and initial dynamic parameters, determine the first control input of the first intelligent agent.

[0103] The control input u of the first intelligent agent i It consists of three parts: obstacle avoidance control input f i βCollision avoidance control input f i g and initial control input f i n The formula is as follows:

[0104]

[0105] Among them, u i f is the first control input. i g Collision avoidance control inputs are used to prevent collisions between multiple agents; f i n f is the initial control input used to move the agent towards the target point; i β To control obstacle avoidance inputs, used to avoid collisions with obstacles.

[0106] Based on the first velocity and first displacement of the first intelligent agent, the second velocity and second displacement of the second intelligent agent, the collision avoidance control input of the first intelligent agent is designed, and the specific formula is as follows:

[0107]

[0108] Among them, f i g For collision avoidance control input, q i p represents the first displacement of the first intelligent agent. i Let q be the first velocity of the first intelligent agent. j p represents the second displacement of the j-th second agent. j Let c1 and c2 be the second velocity of the j-th second agent, and Φ be the control parameters of the first agent. α Let n be the distance function. ij Let a be the displacement vector pointing from the displacement of the first agent to the displacement of the j-th second agent. ij This represents the interaction strength coefficient.

[0109] This formula is used to avoid collisions between multiple agents, primarily considering differences in relative position and velocity. The first term uses the position difference q between neighboring agents. j -q i Calculate the displacement vector n ij Then through the distance function Φ α This measures the influence of neighboring agents on the current agent. This part essentially simulates a "repulsive force"—when agents are too close together, the repulsive force increases, forcing them away from each other to avoid collisions. The second term is the velocity difference p between neighboring agents. j -p iTo adjust the speed inconsistency between them. a ij is an interaction intensity coefficient to ensure that the speed difference does not cause collisions. This is a speed adjustment term that avoids collisions by guiding the speed of an agent closer to that of its neighbor agents.

[0110] In the formula, Φ α (||q j -q i || σ ) is a distance function used to measure the influence of the distance between agents, usually represented by a Gaussian function or other forms. Φ α (z) is a distance function with the distance z as the independent variable:

[0111]

[0112] The distance function is used to measure the mutual force between two agents. It is calculated by the product of a window function ρ h and another potential function Φ(z - d α ). It remains 1 when the distance is close, i.e., full force acts; when the distance is moderate, it gradually weakens as the distance increases; when the distance is too far, the force is 0. This design ensures that an agent only has a significant impact on its closer neighbors. The potential function Φ(z - d α ) represents the intensity of the force, z represents the distance between two agents, and the constant offset d α is a threshold that controls the zero point of the force. By adjusting d α , the minimum allowable distance between agents can be defined.

[0113] Among them, the normalization scale r a and the constant offset d a are calculated as follows respectively:

[0114]

[0115] Among them, the potential function Φ(z) is expressed as follows:

[0116]

[0117] Among them, σ1(z) is an S-shaped function used to smooth the influence of distance on the force, and the constants a, b satisfy 0 < a ≤ b, and c satisfies the following conditions:

[0118]

[0119] This entire function construction ensures that when the distance is small, the interaction between agents is strong, and as the distance increases, the force gradually weakens until it approaches zero.

[0120] Window function ρ h The definition of (z) is:

[0121]

[0122] This is a window function with a smooth transition, used to adjust the range and intensity of interactions between agents. When the distance z is small, i.e., z is within the interval [0, h], the interaction intensity remains at its maximum (i.e., 1). When z is in the transition interval [h, 1], the interaction intensity gradually decreases, smoothly transitioning to zero. When z > 1, the interaction intensity directly becomes zero, indicating that the distance between agents is too large and they no longer interact. This function simulates the smooth decay of interaction forces with distance in reality, ensuring that interactions gradually disappear when agents are far apart, avoiding abrupt changes.

[0123] In one possible implementation, a displacement vector is determined based on the first displacement and the second displacement, and the displacement vector is specifically represented as follows:

[0124]

[0125] Where, n ij Let q be the displacement vector. i Let q be the first displacement of the first intelligent agent. j It represents the second displacement of the j-th second agent.

[0126] n ij This represents the vector pointing from the position of the first agent to the position of the j-th second agent, reflecting the directional relationship between the two agents. The numerator q... j -q i is the displacement difference between agents i and j, representing the direction vector from i to j. The denominator is a normalization factor that ensures the vector calculation takes into account the influence of the distance between agents, so that the length of the direction vector decreases accordingly as the distance increases. This formula can be understood as introducing nonlinear scaling on the basis of traditional vector normalization, adjusting the influence strength of the vector, especially when the distance is very small, to avoid infinite growth.

[0127] Based on the first and second displacements, the interaction strength coefficient is determined using the following formula:

[0128]

[0129] Among them, a ij q is the interaction strength coefficient, used to represent the interaction strength between the first agent i and the j-th second agent. This value varies within the interval [0,1]. j -q i || σ It means (q) j -qi The σ norm q i Let q be the first displacement of the first intelligent agent. j For the second displacement of the j-th second agent, the σ norm is defined as follows:

[0130]

[0131] Where, the constant ε>0, ‖x‖ σ Let ||x|| denote the σ-norm of vector x, and ||x|| denote the length of the vector, where x is a vector of arbitrary dimension. σ It is a norm of nonlinear transformation, suitable for scenarios where the small-scale effects of large-distance transformations need to be considered. This norm is defined by introducing a small parameter ∈ to nonlinearly scale the standard Euclidean distance ||x||. When ∈ is very small, ||x|| σ It is approximately equal to ||x||; when ∈ is large, the norm becomes more sensitive to large distances. When dealing with large distances, this non-linear scaling can more accurately describe the effects of long distances without increasing computational complexity due to excessively large distances.

[0132] In one possible implementation, in a multi-agent system, the motion control of an agent needs to be dynamically adjusted based on its perception of the surrounding environment. In this process, the observation space is also the core basis for the agent's decision-making. The observation space Si includes the following three parts: the agent's own state information, including its current position (first displacement) and current velocity (first velocity). This information describes the agent's basic state in the environment. The agent's observation information of other surrounding agents. This information can include the relative position and velocity with neighboring agents, used to determine potential collision risks or coordinate movement. By integrating the above information, the agent can comprehensively perceive the spatial layout of itself, other agents, and obstacles, thereby making reasonable obstacle avoidance decisions. Based on the first displacement and the first velocity, the observation space is determined, and the observation space is represented as follows:

[0133]

[0134] Among them, s i For the observation space, s i self Let s be the first displacement and the first velocity. i other S represents the observation of the second agent by the first agent. i obstacle This refers to the first intelligent agent's observation of obstacles.

[0135] After acquiring the observation space Si, the first agent determines its velocity change according to the following logic: If an obstacle is detected, the agent calculates a new movement strategy based on the obstacle's position and velocity. This typically involves adjusting the direction or magnitude of the velocity to avoid collisions with obstacles. Specifically, this includes calculating the required avoidance direction based on the obstacle's distance and orientation. Simultaneously, the velocity magnitude is adjusted to ensure the agent does not collide with obstacles during obstacle avoidance while maintaining the target velocity as much as possible. If no obstacles are detected, the agent maintains its original velocity and continues to perform its predetermined task without needing to adjust its velocity.

[0136] The obstacle avoidance control input is adjusted based on the agent's observation of obstacles, specifically according to the following rules: If an obstacle is observed, the obstacle avoidance control input will be equal to the velocity change Δp. i This change is the result of the agent adjusting based on obstacle information, ensuring that the agent can safely avoid obstacles. The formula is expressed as:

[0137]

[0138] If no obstacle is detected: When the agent does not detect an obstacle, the obstacle avoidance control input will be zero, meaning no additional speed adjustment will be made. In this case, the agent simply needs to continue executing its original movement plan. The formula is expressed as:

[0139]

[0140] In one possible implementation, after determining the first control input of the first agent based on motion parameters, obstacle positions, and initial dynamic parameters, the method further includes: acquiring multiple obstacles observed by the first agent and multiple second agents; determining the distance between each obstacle and the first agent; identifying a target obstacle among the multiple obstacles, the target obstacle having the shortest distance to the first agent; and establishing a reward function based on the distance between the target obstacle and the first agent, the reward function being as follows:

[0141]

[0142] Where, r i For the reward function, d i d1 is the distance between the first agent and the target obstacle, μ is the repulsion coefficient, and d0 is the reward adjustment scale.

[0143] Specifically, firstly, it is necessary to obtain the observation results of multiple agents regarding obstacles in the environment. Each agent (including a first agent and multiple second agents) acquires obstacle information. Each agent generates an obstacle list, recording the location information and distances of multiple obstacles. For each observed obstacle, the Euclidean distance between it and the first agent is calculated. Through this process, a list of distances between multiple obstacles and the first agent can be obtained.

[0144] From the list of obstacles, select the obstacle closest to the first agent; this is called the target obstacle. The target obstacle is the obstacle with the smallest distance to the agent and is also the obstacle most likely to affect the agent's obstacle avoidance behavior. By comparing the distances of all obstacles, the target obstacle with the smallest distance is selected.

[0145] In multi-agent obstacle avoidance tasks, the reward function is a core component in designing agent behavior through reinforcement learning. It guides the agent to learn towards the target and continuously adjusts the strategy to achieve better performance. For obstacle avoidance problems, the reward function needs to pay particular attention to the safe distance between the agent and obstacles to avoid collisions. The reward function determines the reward value based on the distance between the agent and the obstacle and is designed in two ways: if the distance between the agent and the obstacle is less than or equal to the critical safe distance d1, the agent will receive a negative reward. The magnitude of the negative reward is the reciprocal of the distance, 1 / d. i The reward is directly proportional to the distance. This means that the closer the distance, the more negative the reward value, i.e., the greater the penalty the agent receives, thus incentivizing the agent to stay away from obstacles as much as possible. The repulsion coefficient determines the agent's sensitivity to obstacle avoidance. The larger the repulsion coefficient, the stronger the agent's incentive to avoid obstacles. d0 is a constant to ensure that even at a large distance (i.e., there is a certain safe distance between the agent and the obstacle), the value of the reward function does not become excessively large.

[0146] The reward function is zero when the distance between the agent and the obstacle is greater than or equal to the critical safe distance d1. This means that within the safe distance, the agent is no longer penalized nor does it receive any additional reward, thus encouraging the agent to move freely while ensuring safety.

[0147] This reward function drives the agent to avoid obstacles and collisions by applying negative rewards within a critical safe distance. This design effectively guides the agent to learn obstacle avoidance behavior within a reinforcement learning framework. Furthermore, by adjusting parameters, the agent's sensitivity to obstacles can be tuned to adapt to different task requirements and environmental complexities. During reinforcement learning, as the agent's understanding of the environment deepens, this reward function gradually guides the agent to learn optimized paths, effectively avoiding obstacles while more efficiently achieving the task objective.

[0148] S150, based on the first control input, perform obstacle avoidance simulation verification on the first intelligent agent.

[0149] In one possible implementation, obstacle avoidance simulation verification is performed on the first intelligent agent based on the first control input, specifically including: constructing a simulation environment, which includes at least two virtual obstacles and a virtual target point; constructing a virtual intelligent agent model based on the first control input; allowing the virtual intelligent agent to interact in the simulation environment; determining whether the virtual intelligent agent can avoid the virtual obstacles and reach the virtual target point; if it is determined that the virtual intelligent agent avoids the virtual obstacles and reaches the virtual target point, then the first control input is determined to be an obstacle avoidance strategy.

[0150] Specifically, a simulation environment is created containing multiple dynamic and static obstacles to increase the complexity of the obstacle avoidance task. A virtual target point is set within this environment; the agent must avoid the obstacles and reach this target point. The target point's location can be fixed or dynamically changing. At least two obstacles are placed to simulate the agent's obstacle avoidance task in a complex environment. The obstacles should have different sizes, shapes, and motion characteristics to increase the difficulty of the agent's perception and decision-making.

[0151] For example, refer to Figure 2 In the XY plane Cartesian coordinate system, a swarm obstacle avoidance problem with N=50 agents is studied. The agents are initially randomly and uniformly distributed within a rectangular region x∈[100,50], y∈[100,100], with an initial velocity of 0. The swarm needs to overcome two circular obstacles centered at (100,50) and (100,100) to reach the target point at coordinates (300,75). Other relevant simulation parameters: ε=0.1, a=b=4.5, c=0, h=0.2, d=16, r=20, e1=0.5, e2=0, c1=c2=1, μ=30, d0=15, d1=12, simulation step size 0.01s, simulation duration 1s.

[0152] A dynamic model of the virtual agent is constructed based on the first control input. The agent obtains information about obstacles in the environment and the location of target points through sensors or perception modules, thereby determining its obstacle avoidance strategy. The agent's motion model determines how its displacement and velocity change according to the control input. The model can employ simplified dynamics or more complex physical simulations (such as considering factors like friction and gravity). (Refer to...) Figure 3 The agent makes real-time decisions based on its observed environmental information (obstacles and target points), generates new control inputs, and executes corresponding motion behaviors. The agent needs to continuously update its state during movement and interact with obstacles and target points in the simulated environment. Dynamic environments in the real world can be simulated by introducing dynamic obstacles (such as moving vehicles or other agents) to test the agent's adaptability in complex environments.

[0153] The simulation system monitors the distance between the agent and obstacles in real time. If the agent maintains a safe distance, obstacle avoidance is considered successful. If the agent collides with an obstacle, obstacle avoidance is considered a failure. (See reference...) Figure 4 Under the premise that the agent successfully avoids obstacles, if it can reach the target point within a set time, the agent is considered to have successfully completed the task. The standard for reaching the target point can be based on the distance between the agent and the target point being less than a certain threshold. If the agent can successfully avoid obstacles and reach the target point in multiple simulation tests, the adopted first control input can be considered an effective obstacle avoidance strategy. The same obstacle avoidance strategy is tested in different scenarios to verify its universality and robustness in various environments. Furthermore, based on the simulation results, the performance of the obstacle avoidance strategy is evaluated, including indicators such as obstacle avoidance success rate, time to reach the target point, and path efficiency. If failures occur in the simulation, the reward function, control input, or algorithm parameters are further adjusted to iteratively optimize the obstacle avoidance strategy.

[0154] This application dynamically acquires and analyzes the initial dynamic parameters of the agent, observes the positions of other agents and obstacles in the real-time observation space, and combines this with the agent's motion parameters to dynamically determine the agent's control input, effectively coping with complex and dynamic environments. The obstacle avoidance effect of this control input is verified in a simulation environment, avoiding static presets and dependencies on the global environment, greatly improving the algorithm's adaptability and flexibility. Furthermore, by adaptively adjusting the agent's behavior, the dependence on the environmental model is reduced, thereby improving the obstacle avoidance performance and practicality of multi-agent clusters in uncertain and dynamic environments.

[0155] This embodiment also discloses a multi-agent cluster obstacle avoidance device, referring to... Figure 5 The device includes an acquisition module 501, a judgment module 502, a processing module 503, and a simulation module 504, wherein:

[0156] The acquisition module 501 is used to acquire the initial dynamic parameters of the intelligent agent cluster, which includes multiple intelligent agents. The initial dynamic parameters include displacement, velocity, and initial control input.

[0157] The judgment module 502 is used to determine multiple second agents within a preset range of the first agent based on the initial dynamic parameters, wherein the first agent is any one of the multiple agents.

[0158] The acquisition module 501 is used to acquire the motion parameters of the first intelligent agent and the observed obstacle positions.

[0159] The processing module 503 is used to determine the first control input of the first intelligent agent based on motion parameters, obstacle positions and initial dynamic parameters.

[0160] The simulation module 504 is used to perform obstacle avoidance simulation verification on the first intelligent agent based on the first control input.

[0161] In one possible implementation, the processing module 503 is configured to determine the collision avoidance control input of the first intelligent agent based on the first velocity of the first intelligent agent, the first displacement of the first intelligent agent, the second velocity of the second intelligent agent, and the second displacement of the second intelligent agent. The collision avoidance control input is represented as follows:

[0162]

[0163] Among them, f i g For collision avoidance control input, q i p represents the first displacement of the first intelligent agent. i Let q be the first velocity of the first intelligent agent. j p represents the second displacement of the j-th second agent. j Let c1 and c2 be the second velocity of the j-th second agent, and c1 and c2 be the control parameters of the first agent. ij Let a be the displacement vector pointing from the displacement of the first agent to the displacement of the j-th second agent. ij Φ is the interaction strength coefficient. α This is the distance function.

[0164] The acquisition module 501 is used to acquire the target point position and target point velocity of the intelligent agent cluster.

[0165] Processing module 503 is used to determine the initial control input based on the target point position and the target point velocity. The initial control input is represented as follows:

[0166] f i n =-e1(q i -q target )-e2(pi -p target )

[0167] Among them, f i n As the initial control input, q i For the first displacement, p i As the first velocity, q target p is the target point location. target e1 represents the target point velocity, and e2 represents the control parameters.

[0168] Processing module 503 is used to determine a first control input based on the collision avoidance control input and the initial control input. The first control input is represented as follows:

[0169]

[0170] Among them, u i f is the first control input. i g To avoid collisions between multiple intelligent agents, the collision avoidance control input, f i n f is the initial control input. i β To control obstacle avoidance input.

[0171] In one possible implementation, the processing module 503 is used to determine the observation space based on the first displacement and the first velocity, the observation space being represented as follows:

[0172]

[0173] Among them, s i For the observation space, s i self Let s be the first displacement and the first velocity. i other S represents the observation of the second agent by the first agent. i obstacle This refers to the first intelligent agent's observation of obstacles.

[0174] Processing module 503 is used to determine the control obstacle avoidance input based on the observation space. The control obstacle avoidance input is represented as follows:

[0175]

[0176] Among them, f i β To control the obstacle avoidance input, Δp i This represents the change in velocity.

[0177] In one possible implementation, the processing module 503 is used to determine a displacement vector based on the first displacement and the second displacement, wherein the displacement vector is specifically represented as follows:

[0178]

[0179] Where, n ij Let q be the displacement vector. i Let q be the first displacement of the first intelligent agent. j This represents the second displacement of the j-th second agent;

[0180] Processing module 503 is used to determine the interaction strength coefficient based on the first displacement and the second displacement, specifically calculated using the following formula:

[0181]

[0182] Among them, a ij Let q be the interaction strength coefficient. j -q i || σ It means (q) j -q i The σ norm of q i Let q be the first displacement of the first intelligent agent. j For the second displacement of the j-th second agent, the σ norm is defined as follows:

[0183]

[0184] Where, the constant ε>0, ‖x‖ σ Let σ represent the σ norm of vector x, and ‖x‖ represent the length of the vector, where x is a vector of arbitrary dimensions.

[0185] In one possible implementation, the acquisition module 501 is used to acquire multiple obstacles observed by the first intelligent agent and multiple second intelligent agents.

[0186] The judgment module 502 is used to determine the distance between each obstacle and the first intelligent agent.

[0187] The judgment module 502 is used to determine the target obstacle among multiple obstacles, and the target obstacle is closest to the first intelligent agent.

[0188] Processing module 503 is used to establish a reward function based on the distance between the target obstacle and the first intelligent agent. The reward function is as follows:

[0189]

[0190] Where, r i For the reward function, d id1 is the distance between the first agent and the target obstacle, μ is the repulsion coefficient, and d0 is the reward adjustment scale.

[0191] In one possible implementation, the processing module 503 is used to determine a distance function based on the distance between the first agent and the second agent, specifically expressed by the following formula:

[0192]

[0193] Where, Φ α (z) is a distance function with distance z as the independent variable, z = ||q j -q i ||,r a For the normalization scale, d a This is a constant offset.

[0194] In one possible implementation, simulation module 504 is used to construct a simulation environment, which includes at least two virtual obstacles and virtual target points.

[0195] Simulation module 504 is used to construct a virtual intelligent agent model based on the first control input.

[0196] Simulation module 504 is used to enable virtual intelligent agents to interact in a simulation environment.

[0197] The judgment module 502 is used to determine whether the virtual intelligent agent can avoid virtual obstacles and reach the virtual target point.

[0198] The judgment module 502 is used to determine the first control input as an obstacle avoidance strategy if it is determined that the virtual intelligent agent avoids virtual obstacles and reaches the virtual target point.

[0199] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0200] This embodiment also discloses an electronic device, as shown in the reference. Figure 6 The electronic device may include: at least one processor 601, at least one communication bus 602, user interface 603, network interface 604, and at least one memory 605.

[0201] The communication bus 602 is used to enable communication between these components.

[0202] The user interface 603 may include a display screen and a camera. Optionally, the user interface 603 may also include a standard wired interface and a wireless interface.

[0203] The network interface 604 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0204] The processor 601 may include one or more processing cores. The processor 601 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 605, and by calling data stored in the memory 605. Optionally, the processor 601 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 601 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content to be displayed on the screen; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the processor 601 and may be implemented as a separate chip.

[0205] The memory 605 may include random access memory (RAM) or read-only memory. Optionally, the memory may include a non-transitory computer-readable storage medium. The memory 605 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 605 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 605 may also be at least one storage device located remotely from the aforementioned processor 601. As a computer storage medium, the memory 605 may include an operating system, a network communication module, a user interface 603 module, and an application program for a multi-agent swarm obstacle avoidance method.

[0206] exist Figure 6 In the electronic device shown, the user interface 603 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 601 can be used to call an application program stored in the memory 605 for a multi-agent cluster obstacle avoidance method. When executed by one or more processors 601, the electronic device executes one or more methods as described in the above embodiments.

[0207] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0208] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0209] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be through some service interfaces; indirect couplings or communication connections between apparatuses or units may be electrical or other forms.

[0210] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0211] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0212] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory 605 and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory 605 includes various media capable of storing program code, such as a USB flash drive, external hard drive, magnetic disk, or optical disk.

[0213] This application also discloses a computer-readable storage medium storing instructions. When executed by one or more processors 601, these instructions cause an electronic device to perform one or more of the methods described in the above embodiments.

[0214] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of other embodiments of this disclosure upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A multi-agent swarm obstacle avoidance method, characterized in that, The method includes: The initial dynamic parameters of the intelligent agent cluster, which comprises multiple intelligent agents, are obtained. These initial dynamic parameters include displacement, velocity, and initial control input. Based on the initial dynamic parameters, a plurality of second agents within a preset range of the first agent are determined, wherein the first agent is any one of the plurality of agents. Obtain the motion parameters of the first intelligent agent and the observed positions of obstacles; Based on the motion parameters, the obstacle position, and the initial dynamic parameters, the first control input of the first intelligent agent is determined; Based on the first control input, the first intelligent agent is subjected to obstacle avoidance simulation verification; The determination of the first control input for the first intelligent agent based on the motion parameters, the obstacle position, and the initial dynamic parameters specifically includes: Based on the first velocity of the first agent, the first displacement of the first agent, the second velocity of the second agent, and the second displacement of the second agent, the collision avoidance control input of the first agent is determined, and the collision avoidance control input is expressed as follows: ； Among them, f i g q is the collision avoidance control input. i p represents the first displacement of the first intelligent agent. i Let q be the first velocity of the first intelligent agent. j p represents the second displacement of the j-th second agent. j Let c1 and c2 be the second velocity of the j-th second agent, and c1 and c2 be the control parameters of the first agent. ij Let a be the displacement vector from the displacement of the first agent to the displacement of the j-th second agent. ij Φ is the interaction strength coefficient. α It is a distance function; Obtain the target point position and target point velocity of the intelligent agent cluster; Based on the target point position and the target point velocity, the initial control input is determined as follows: ； Among them, f i n q is the initial control input. i For the first displacement, p i Let q be the first velocity. target p is the location of the target point. target The target point velocity is given by e1 and e2, which are control parameters. Based on the collision avoidance control input and the initial control input, the first control input is determined, and the first control input is represented as follows: ； Among them, u i f is the first control input. i g To avoid collisions between multiple intelligent agents, f is the collision avoidance control input. i n f is the initial control input. i β To control obstacle avoidance input; The step of determining the first control input based on the obstacle avoidance control input further includes: Based on the first displacement and the first velocity, the observation space is determined as follows: ； Among them, s i For the observation space, s i self Let s be the first displacement and the first velocity. i other S represents the observation of the first agent on the second agent. i obstacle This refers to the first intelligent agent's observation of obstacles; Based on the observation space, the control obstacle avoidance input is determined, and the control obstacle avoidance input is represented as follows: ； Among them, f i β Δp is the control input for obstacle avoidance. i This represents the change in velocity.

2. The multi-agent swarm obstacle avoidance method according to claim 1, characterized in that, The step of determining the collision avoidance control input of the first intelligent agent based on the first velocity, the first displacement of the first intelligent agent, the second velocity of the second intelligent agent, and the second displacement of the second intelligent agent further includes: Based on the first displacement and the second displacement, the displacement vector is determined, and the displacement vector is specifically represented as follows: ； Where, n ij Let q be the displacement vector. i Let q be the first displacement of the first intelligent agent. j This represents the second displacement of the j-th second agent; Based on the first displacement and the second displacement, the interaction strength coefficient is determined, specifically calculated using the following formula: ； Among them, a ij Let q be the interaction strength coefficient. j -q i || σ It means (q) j -q i )of Norm, q i Let q be the first displacement of the first intelligent agent. j For the second displacement of the j-th second agent, the σ norm is defined as follows: ； Where, the constant ε>0, ‖x‖ σ Let σ represent the σ norm of vector x, and ‖x‖ represent the length of the vector, where x is a vector of arbitrary dimensions.

3. The multi-agent swarm obstacle avoidance method according to claim 1, characterized in that, After determining the first control input of the first agent based on the motion parameters, the obstacle position, and the initial dynamic parameters, the method further includes: Acquire multiple obstacles observed by the first intelligent agent and multiple second intelligent agents; Determine the distance between each obstacle and the first intelligent agent; Identify a target obstacle among multiple obstacles, the target obstacle being the closest to the first agent; A reward function is established based on the distance between the target obstacle and the first agent, and the reward function is as follows: ； Where, r i Let d be the reward function. i d1 is the distance between the first agent and the target obstacle, μ is the repulsion coefficient, and d0 is the reward adjustment scale.

4. The multi-agent swarm obstacle avoidance method according to claim 3, characterized in that, The distance function is determined based on the distance between the first agent and the second agent, and is specifically expressed by the following formula: ； Where, Φ α (z) is a distance function with distance z as the independent variable, r a For the normalization scale, d a This is a constant offset.

5. The multi-agent swarm obstacle avoidance method according to claim 1, characterized in that, The step of performing obstacle avoidance simulation verification on the first intelligent agent based on the first control input specifically includes: Construct a simulation environment, which includes at least two virtual obstacles and virtual target points; A virtual intelligent agent model is constructed based on the first control input; The virtual intelligent agent interacts within the simulation environment; Determine whether the virtual agent can avoid the virtual obstacles and reach the virtual target point; If it is determined that the virtual agent avoids the virtual obstacle and reaches the virtual target point, then the first control input is determined to be an obstacle avoidance strategy.

6. A multi-agent swarm obstacle avoidance device, characterized in that, The apparatus is used to perform the method as described in any one of claims 1-5, the apparatus comprising an acquisition module (501), a judgment module (502), a processing module (503), and a simulation module (504), wherein: The acquisition module (501) is used to acquire the initial dynamic parameters of the intelligent agent cluster, which includes multiple intelligent agents. The initial dynamic parameters include displacement, velocity, and initial control input. The judgment module (502) is used to determine a plurality of second agents within a preset range of the first agent based on the initial dynamic parameters, wherein the first agent is any one of the plurality of agents; The acquisition module (501) is used to acquire the motion parameters of the first intelligent agent and the observed obstacle positions; The processing module (503) is used to determine the first control input of the first intelligent agent based on the motion parameters, the obstacle position and the initial dynamic parameters; The simulation module (504) is used to perform obstacle avoidance simulation verification on the first intelligent agent according to the first control input.

7. An electronic device, characterized in that, The device includes a processor (601), a communication bus (602), a memory (605), a user interface (603), a network interface (604), and a memory (605). The memory (605) is used to store instructions. The user interface (603) and the network interface (604) are both used to communicate with other devices. The processor (601) is used to execute the instructions stored in the memory (605) to cause the electronic device to perform the method as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1-5.

Images