A method, system, equipment, and medium for multi-target cooperative tracking using swarm robots for highway inspection.
By using a density-interaction-based swarm control model and an enhanced particle swarm optimization algorithm, the problem of multi-target tracking in complex environments for swarm robots was solved, achieving self-organizing, safe, and robust multi-target tracking performance.
Patent Information
- Application Number
- CN202611372976.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-07
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2046-09-07
AI Technical Summary
Existing swarm robot technologies struggle to achieve multi-target tracking in complex environments, especially in dynamic environments where it is difficult to guarantee a safe distance and collision-free safety between the robot and the target point. It is also difficult to adjust control parameters in real time for rapidly moving dynamic target points, making robot resource allocation challenging in multi-target tracking scenarios.
A density-interaction-based swarm control model is adopted, combined with an enhanced particle swarm optimization algorithm, to dynamically optimize control parameters. Through density interaction terms, collision avoidance terms, and obstacle avoidance terms, the robot can self-organize and track multiple moving targets, ensuring safe distance and obstacle avoidance capabilities.
It achieves efficient and safe multi-target collaborative tracking in complex environments, adaptively adjusts parameters to respond to environmental changes, improves system robustness and tracking efficiency, and avoids collisions between the robot and targets and obstacles.
Smart Images

Figure CN122902596B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of swarm robot control technology, and in particular to a method, system, device and medium for multi-target cooperative tracking of swarm robots for highway inspection. Background Technology
[0002] Swarm robot cooperative target tracking refers to a group of intelligent, autonomous mobile robots that, by sensing the dynamic environment around them, interact and coordinate with neighboring robots within their sensing range, autonomously make decisions and plan their movements to adjust their relative positions to multiple target objects, thereby achieving the encirclement and capture of a specific target. In this system, each robot autonomously perceives and makes its own decisions, employing a distributed control strategy.
[0003] Multi-target tracking control for swarm robots can be applied to multi-area monitoring, such as multiple drones collaboratively monitoring multiple moving targets; it can also be applied to environmental monitoring, such as monitoring multiple key parameter points in marine or forest environments; and search and rescue missions, such as collaboratively rescuing multiple trapped individuals at disaster sites. Despite significant progress in collaborative tasks, swarm robots still face considerable challenges in achieving multi-target tracking in complex environments. Traditional methods typically rely on global task allocation or explicit target allocation strategies, which often fail in dynamic environments. For example, existing tracking algorithms based on artificial potential fields are prone to getting trapped in local minima when encountering obstacles, leading to robot oscillations or ineffective target tracking; methods based on pairwise interaction models struggle to automatically split and merge swarms in multi-target scenarios, failing to form efficient tracking configurations.
[0004] Although swarm robot technology has made some progress in target tracking in recent years, existing methods still face some key challenges: (1) It is difficult to ensure a safe distance between the robot and the target point and the collision-free safety between robots during dynamic tracking; (2) For fast-moving dynamic target points, there is a lack of effective mechanisms to adjust control parameters in real time to respond to changes in the target's motion state, resulting in low tracking efficiency; (3) In multi-target tracking scenarios, it is difficult to achieve self-organized allocation of robot resources without relying on explicit task allocation.
[0005] Therefore, it is necessary to provide a multi-target cooperative tracking method for swarm robots used in highway inspection to solve the above problems. Summary of the Invention
[0006] The purpose of this application is to provide a method, system, device, and medium for collaborative tracking of multiple moving targets by swarm robots for highway inspection, which can efficiently, safely, and self-organize the collaborative tracking of multiple moving targets.
[0007] To achieve the above objectives, this application provides the following solution: Firstly, this application provides a multi-target cooperative tracking method for swarm robots used in highway inspection, the method comprising: Acquire the local perception data of each robot in the swarm robot tracking system at the current moment; the local perception data includes: the position and velocity of all robots themselves, the position of all neighboring robots within the perception radius of each robot, the position of all moving targets, and the geometric information of all obstacles; Establish an initial expression for a density-interaction-based swarm control model; the density-interaction-based swarm control model is a function of control parameters, inter-robot interaction terms, robot-target interaction terms, inter-robot collision avoidance terms, target collision prevention terms, obstacle collision avoidance terms, and speed adjustment terms; An enhanced particle swarm optimization algorithm is used to dynamically optimize the control parameters in a density-interaction-based swarm control model, obtaining the optimized values of the control parameters at the current time step. Substituting the optimized values of the control parameters at the current moment into the initial expression of the density-interaction-based swarm control model, we obtain the objective expression of the density-interaction-based swarm control model at the current moment. Taking any robot in the swarm as the current robot, based on the local perception data of the current robot at the current moment and the objective expression of the swarm control model based on density interaction at the current moment, the estimated value of the control quantity of the current robot at the current moment is determined, and then the estimated value of the control quantity of all robots at the current moment is determined. Based on the estimated control values of all robots at the current moment, the velocity and position of all robots at the next moment are updated to enable tracking of multiple moving targets while avoiding obstacles.
[0008] In one implementation, the initial expression for the density-interaction-based swarm control model is: ; in, For robots The estimated value of the control quantity; For robots Inter-robot interaction items; For robots Interaction items between the robot and the target; For robots Robot collision avoidance; For robots Target collision prevention items; For robots Obstacle avoidance items; For robots Speed adjustment option.
[0009] In one embodiment, the control parameters include: inter-robot density interaction strength coefficient, robot-target density interaction strength coefficient, robot-robot collision avoidance repulsion strength coefficient, robot-target point repulsion strength coefficient, robot-obstacle repulsion strength coefficient, and reference density.
[0010] In one implementation, the expression for the robot-to-robot interaction term is: ; in, This represents the density interaction intensity coefficient between robots. For robots Local spatial density; For reference density; For robots The gradient operator; The expression for the interaction term between the robot and the target is: ; in, The robot-target density interaction intensity coefficient; is the cubic spline kernel function; k is the number of the moving target; For robots A set of moving targets within the perception radius; For the k-th moving target to the robot The relative position vector; For the k-th moving target to the robot The distance; For sensing radius; The expression for the robot collision avoidance term is: ; in, The number assigned to the neighboring robot; For robots A collection of neighboring robots within the perception radius; The robot-to-robot collision avoidance repulsion strength coefficient; For robots Neighbor Robot The relative position vectors between them; The expression for the target collision prevention term is: ; in, The robot-target point repulsion strength coefficient; The expression for the obstacle collision avoidance term is: ; in, For robots A set of obstacles within the perception radius; The robot-obstacle repulsion strength coefficient; The table shows the data from the robot. To the obstacle The shortest vector on the boundary; The expression for the speed adjustment term is: ; in, This is the acceleration coefficient; It is the damping factor; For robots The speed.
[0011] In one embodiment, an enhanced particle swarm optimization algorithm is used to dynamically optimize the control parameters in a density-interaction-based swarm control model to obtain the optimized values of the control parameters at the current time step, specifically including: Construct a parameter vector to be optimized, consisting of the density interaction strength coefficient between robots, the density interaction strength coefficient between robots and targets, the collision avoidance and repulsion strength coefficient between robots and robots, the repulsion strength coefficient between robots and targets, the repulsion strength coefficient between robots and obstacles, and the reference density. An enhanced particle swarm optimization algorithm is used to optimize the parameter vector to be optimized in real time, with the goal of maximizing the overall performance index of the swarm robot tracking system. The comprehensive performance indicators include: position uniformity index, numerical distribution index, connectivity index, collision avoidance index between robots, collision avoidance index between robot and target, and collision avoidance index between robot and obstacle.
[0012] In one embodiment, the expression for the comprehensive performance index is: ; in, The comprehensive performance index function represents the overall performance index function. Time and system state are The parameter vector to be optimized is Overall performance score at the time; for The status of the swarm robot tracking system at any given moment; for The vector of parameters to be optimized at time step; for The calculated value of the positional uniformity index at any given time; for Calculated values of the numerical distribution index at any given time; for The calculated value of the connectivity index at any given time; for The calculated value of the collision avoidance index between robots at any given time; for The calculated value of the collision avoidance index between the robot and the target at any given time; for The calculated value of the collision avoidance index between the robot and obstacles at any given time.
[0013] In one embodiment, the expression for the location uniformity index is: ; in, for The calculated value of the positional uniformity index at any given time; The total number of moving targets; To move the target k A collection of robots that interact with each other; With moving targets k The number of robots that interact; For robots Location; For the first k The location of the moving target; The expression for the numerical distribution index is: ; in, for The calculated value of the numerical distribution index at time. This represents the expected number of robots for each moving target; It represents the actual number of robots orbiting the moving target k; N is the total number of robots. The expression for the connectivity index is: ; in, for The calculated value of the connectivity index at any given time; The number of robots that interact with at least one moving target; The expression for the collision avoidance index between robots is: ; in, for The calculated value of the collision avoidance index between robots at any given time; For smooth transition functions; For any pair of robots in The minimum distance at any given moment; It is the desired separation distance between the robots; It is the minimum safe distance between the robots; The expression for the collision avoidance index between the robot and the target is: ; in, for The calculated value of the collision avoidance index between the robot and the moving target at any given time; For any robot and any moving target in The minimum distance at any given moment; The desired separation distance between the robot and the target; The critical safety distance between the robot and the target; The expression for the collision avoidance index between the robot and obstacles is: ; in, Let t be the calculated value of the collision avoidance index between the robot and the obstacle; For any robot to reach any obstacle boundary in The shortest distance at any given moment; The desired separation distance between the robot and the obstacle; This is the minimum safe distance between the robot and the obstacle.
[0014] Secondly, this application provides a multi-target cooperative tracking system for swarm robots used in highway inspection. The system is used to implement the aforementioned multi-target cooperative tracking method for highway inspection. The multi-target cooperative tracking system for swarm robots used in highway inspection includes: The data acquisition unit is used to acquire the local perception data of each robot in the swarm robot tracking system at the current moment. The local perception data includes: the position and velocity of all robots themselves, the position of all neighboring robots within the perception radius of each robot, the position of all moving targets, and the geometric information of all obstacles. The initial model building unit is used to establish the initial expression of the density-interaction-based swarm control model; the density-interaction-based swarm control model is a function of control parameters, robot-to-robot interaction terms, robot-to-target interaction terms, robot-to-robot collision avoidance terms, target collision prevention terms, obstacle collision avoidance terms, and speed adjustment terms. The control parameter optimization unit is used to dynamically optimize the control parameters in the density interaction-based swarm control model using an enhanced particle swarm optimization algorithm to obtain the optimized values of the control parameters at the current time. The target model determination unit is used to substitute the optimized values of the control parameters at the current time into the initial expression of the density-interaction-based swarm control model to obtain the target expression of the density-interaction-based swarm control model at the current time. The control quantity estimation unit is used to take any robot in the swarm as the current robot, and based on the local perception data of the current robot at the current time and the objective expression of the density-interaction-based swarm control model at the current time, determine the estimated value of the control quantity of the current robot at the current time, and then determine the estimated value of the control quantity of all robots at the current time. The speed and position update unit is used to update the speed and position of all robots in the next moment based on the estimated control values of all robots at the current moment, so as to achieve obstacle avoidance while tracking multiple moving targets.
[0015] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method for multi-target cooperative tracking of swarm robots for highway inspection.
[0016] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for multi-target cooperative tracking of swarm robots for highway inspection.
[0017] According to the specific embodiments provided in this application, this application has the following technical effects: This application discloses a method, system, device, and medium for multi-target cooperative tracking of swarm robots for highway inspection. First, by constructing a density-based interaction-based swarm control model and introducing specially designed obstacle avoidance and target approach terms, this application effectively solves the technical challenge of "effectively allocating robot resources among multiple targets without explicit task assignment." This mechanism enables robots to autonomously navigate around obstacles while maintaining a safe distance from targets and other robots. The density-based interaction design not only encompasses simple pairwise interactions but also incorporates swarm spatial distribution characteristics, thereby achieving adaptive swarm cooperative behavior and addressing the technical requirement of "simultaneously maintaining target tracking and obstacle avoidance in an obstacle environment." Second, this application employs an enhanced particle swarm optimization algorithm to construct a dynamic parameter optimization framework. This framework continuously adjusts swarm control parameters according to environmental changes, effectively balancing competing objectives such as configuration quality, obstacle avoidance capability, and target tracking efficiency, solving the technical problem of "dynamically optimizing control parameters to respond to environmental changes." Through this adaptive optimization mechanism, the system can maintain stable performance under various noise levels and swarm sizes, thereby improving system robustness. Attached Figure Description
[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0019] Figure 1 This is a schematic flowchart of a multi-target cooperative tracking method for highway inspection using swarm robots, provided in an embodiment of this application. Figure 2 This is a schematic diagram illustrating the effects of different functional items provided in an embodiment of this application; Figure 3 This is a schematic diagram of a multi-target tracking framework based on dynamic optimization provided in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation
[0020] The technical solutions of the embodiments of this application 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. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0021] This application aims to address the key challenges of multi-target tracking in swarm robots in complex obstacle environments. Specifically, this application addresses the following technical problems: 1) how to effectively allocate robot resources among multiple targets without using explicit task assignment; 2) how to simultaneously maintain target tracking and obstacle avoidance in obstacle environments; 3) how to dynamically optimize control parameters to respond to environmental changes and balance configuration quality, obstacle avoidance capability, and tracking efficiency; and 4) how to ensure the robustness of the system under various noise levels and swarm sizes.
[0022] To address this, this application proposes a multi-target cooperative tracking method for swarm robots used in highway inspection. Based on a dynamically optimized density-interactive swarm control framework, it enhances the swarm's ability to track multiple dynamic targets. This method enables the robot swarm to self-organize and form a tracking configuration without centralized control, adaptively adjusting parameters as the environment dynamically changes. This ensures effective tracking of multiple moving targets while maintaining a balanced distribution and uniformity of robot numbers and positions around each target, and avoiding collisions with other robots, targets, and obstacles in the environment.
[0023] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0024] In one exemplary embodiment, such as Figure 1 As shown, a multi-target cooperative tracking method for swarm robots used in highway inspection is provided, including the following steps: Wherein: Step S1: Obtain the local perception data of each robot in the swarm robot tracking system at the current moment; the local perception data includes: the position and velocity of all robots themselves, the position of all neighboring robots within the perception radius of each robot, the position of all moving targets (also known as target points), and the geometric information of all obstacles.
[0025] Specifically, consider a workspace There is A robot A moving target (also called a target point) and There are [number] obstacles. The sensing radius of each robot is [radius]. And follows double integral dynamics: , ,in They are robots Position, velocity, and control input; the position and velocity of the moving target are respectively expressed as... and ,in The obstacle is modeled as a circle, and its geometric information includes: center and radius ,in This allows the robot to pinpoint its location. The objective of this application is to design a distributed control algorithm that enables the robot to track multiple moving targets in a dynamic environment while ensuring that collisions are avoided between the robot, the targets, and obstacles.
[0026] Step S2: Establish the initial expression for the density-interaction-based swarm control model; the density-interaction-based swarm control model is a function of control parameters, robot-to-robot interaction terms, robot-to-target interaction terms, robot-to-robot collision avoidance terms, target collision prevention terms, obstacle collision avoidance terms, and speed adjustment terms. The robot-to-target interaction terms are the interaction terms between the robot and the moving target, and the target collision prevention terms are the collision prevention terms between moving targets.
[0027] Specifically, this application designs a Dynamically Optimized Density Model (DODM) for multi-robot collaborative target tracking. This model combines a density-interaction-based swarm control model with a dynamic parameter optimization framework, enabling the robot swarm to self-organize and form a tracking configuration without centralized control, ensuring effective encirclement of multiple moving targets while avoiding collisions with other robots, targets, and obstacles in the environment.
[0028] As an optional implementation, in step S2, the initial expression of the density-interaction-based group control model is: (1) The initial expression of the density-interaction-based group control model includes multiple action terms. For robots The estimated value of the control quantity; For robots Inter-robot interaction terms, controlling density-based coordination of neighboring robots; For robots The interaction items between the robot and the target guide the robot to move toward the target and maintain an appropriate distance; For robots The robot collision avoidance system prevents collisions between robots during movement; For robots The target collision prevention item creates a safety buffer during the tracking process; For robots The obstacle avoidance features enable the robot to navigate in environments with obstacles; For robots The speed adjustment setting maintains a consistent robot speed. Figure 2 The effects of different functions are shown. Figure 2 This illustrates the impact of different interactions and collision avoidance strategies on group tracking behavior. Figure 2 Small and medium-sized colored circles represent individual robots, large red circles represent moving targets, and gray circles represent obstacles.
[0029] As an optional implementation, the control parameters include: inter-robot density interaction strength coefficient, robot-target density interaction strength coefficient, robot-robot collision avoidance repulsion strength coefficient, robot-target point repulsion strength coefficient, robot-obstacle repulsion strength coefficient, and reference density.
[0030] As an optional implementation, in step S2, the expression for the robot-to-robot interaction item is: (2) in, This represents the density interaction intensity coefficient between robots. ; For robots Local spatial density; For reference density, This defines the spatial distribution of group expectations; For robots The gradient operator. The inter-robot interaction term creates a dynamic equilibrium: when... When, the robot stays away from high-density areas; when As the robot moves toward a higher density area, it eventually converges to a desired spatial configuration, in which... .
[0031] Specifically, density-based interaction mechanisms are a group coordination method that goes beyond simple pairwise interactions, enabling cohesive group behavior while maintaining appropriate robot spacing. The density interaction terms between robots... The determination process is as follows: For robots First, calculate its local spatial density. It is obtained by integrating contributions from neighboring robots, and the calculation formula is as follows: (3) in, Represents robots A collection of neighboring robots within the perception radius; It is a relative position vector. For the first The location of each robot; This is a cubic spline kernel function that provides a smooth, continuous local density estimate that decreases with increasing distance, allowing for more subtle spatial interactions between robots. The kernel function is defined as follows: (4) robot density gradient at location The calculation formula is as follows: (5) In step S2, the expression for the interaction term between the robot and the target is: (6) in, The robot-target density interaction intensity coefficient. ; It is a cubic spline kernel function; k The number of the moving target; For robots A set of moving targets within the perception radius. ; For the first k A moving target to the robot The relative position vector, ; For the first k A moving target to the robot The distance.
[0032] Specifically, in order to achieve effective target tracking, the robot must be attracted to the vicinity of the moving target while maintaining an appropriate configuration. The target attraction mechanism utilizes the concept of density to create an adaptive encirclement pattern, namely the robot-target interaction term as shown in Equation (6).
[0033] In step S2, the expression for the robot collision avoidance term is: (7) in, The number assigned to the neighboring robot; For robots A collection of neighboring robots within the perception radius; The robot-to-robot collision avoidance repulsion strength coefficient. ; For robots Neighbor Robot The relative position vectors between them.
[0034] Specifically, while density-based interactions facilitate appropriate spacing, explicit collision avoidance mechanisms are crucial for ensuring safety, especially in rapid maneuvers or confined spaces. Therefore, this is achieved through a repulsive force term, resulting in the inter-robot collision avoidance term shown in Equation (7). The inverse square law formula corresponding to the inter-robot collision avoidance term creates a repulsive force that increases rapidly as robots approach each other, effectively preventing collisions while minimizing the impact at larger distances.
[0035] In step S2, the expression for the target collision prevention term is: (8) in, The robot-target point repulsion strength coefficient. .
[0036] During tracking, the robot must approach the moving target but not make direct contact. Robot-target collision prevention systems can meet this requirement.
[0037] In step S2, the expression for the obstacle collision avoidance term is: (9) in, For robots A set of obstacles within the perception radius; The robot-obstacle repulsion strength coefficient. ; The table shows the data from the robot. To the obstacle The shortest vector on the boundary. When the robot is outside the obstacle ( When ), this vector is calculated as ,in It is the vector from the center of the obstacle to the robot.
[0038] Specifically, a key enhancement to the model is a dedicated obstacle avoidance mechanism that enables the robot to navigate robustly in complex environments without disrupting its tracking behavior. Therefore, an obstacle avoidance term as shown in Equation (9) was designed.
[0039] In step S2, the expression for the speed regulation term (also known as the self-driving term) is: (10) in, This is the acceleration coefficient; It is the damping factor; For robots The speed.
[0040] Specifically, the self-driving term enables the robot to maintain a consistent desired speed. To ensure stable group dynamics and prevent unrealistic movement patterns, [the following is set up / adjusted]. .
[0041] Step S3: The enhanced particle swarm optimization algorithm is used to dynamically optimize the control parameters in the density interaction-based swarm control model to obtain the optimized values of the control parameters at the current time.
[0042] As an optional implementation, step S3 specifically includes: Step S31: Construct a parameter vector to be optimized, consisting of the density interaction strength coefficient between robots, the density interaction strength coefficient between robots and targets, the collision avoidance and repulsion strength coefficient between robots and robots, the repulsion strength coefficient between robots and targets, the repulsion strength coefficient between robots and obstacles, and the reference density.
[0043] Specifically, through extensive analysis of population dynamics, six key control parameters that significantly affect tracking performance were identified: (11) The aggregation and separation behavior among robots is caused by Control, which regulates density-based interactions between neighboring robots; target tracking and tracking behavior are controlled by... Control influences the intensity with which a robot is attracted to a target; collision prevention between robots is achieved through... Management ensures a safe distance during movement; the safety buffer zone around the target is maintained by... Establishment, preventing direct contact while maintaining close distance; obstacle avoidance capability from The decision enables the robot to maneuver in complex environments; finally, A reference density is defined to establish the optimal spacing throughout the entire group configuration, which determines the final enclosure mode.
[0044] Step S32: An enhanced particle swarm optimization algorithm is used to optimize the parameter vector to be optimized in real time, with the goal of maximizing the comprehensive performance index of the swarm robot tracking system. The comprehensive performance index includes: position uniformity index, numerical distribution index, connectivity index, collision avoidance index between robots, collision avoidance index between robots and the target, and collision avoidance index between robots and obstacles.
[0045] As an optional implementation method, the expression for the comprehensive performance index is: (12) in, The comprehensive performance index function represents the overall performance index function. Time and system state are The parameter vector to be optimized is Overall performance score at the time; for The status of the swarm robot tracking system at any given moment; for The vector of parameters to be optimized at time step; for The calculated value of the position uniformity index at any given time is used to measure the uniformity of the robot's distribution around the target; for The calculated value of the numerical distribution index at time is used to evaluate whether each target has the optimal number of surrounding robots, and to assess the degree of matching between the actual robot allocation and the ideal distribution. for The calculated value of the connectivity index at any given time is used to quantify the proportion of robots actively participating in target tracking; for The calculated value of the collision avoidance index between robots at any given time is used to evaluate the collision avoidance performance between robots; for The calculated value of the collision avoidance index between the robot and the target at any given time is used to evaluate the collision avoidance performance between the robot and the target. for The calculated collision avoidance index between the robot and obstacles at any given time is used to assess the robot's collision avoidance performance. Safety index ( The collision avoidance performance is evaluated by measuring the minimum distance and the desired safety threshold.
[0046] Specifically, a comprehensive fitness function dynamic optimization control parameter was designed to serve as a comprehensive performance index for evaluating the tracking performance of the swarm robot tracking system, as shown in Equation (12).
[0047] As an optional implementation method, the expression for the location uniformity index is: (13) in, for The calculated value of the positional uniformity index at any given time; The total number of moving targets; To move the target A collection of robots that interact with each other; With moving targets The number of robots interacting with the target; this metric reaches its maximum value of 1 when the robots are perfectly distributed around the target. For robots i Location; Let be the position of the k-th moving target.
[0048] The expression for the numerical distribution index is: (14) in, for The calculated value of the numerical distribution index at time. This represents the expected number of robots for each moving target; It represents the actual number of robots orbiting the moving target k; N is the total number of robots.
[0049] The expression for the connectivity index is: (15) in, for The calculated value of the connectivity index at any given time; The number of robots that interact with at least one moving target.
[0050] The expression for the collision avoidance index between robots is: (16) in, for The calculated value of the collision avoidance index between robots at any given time, which evaluates the minimum distance between any two robots; For smooth transition functions; For any pair of robots in The minimum distance at any given moment; It is the desired separation distance between the robots; It is the desired minimum safe distance between robots; here, , .
[0051] The expression for the collision avoidance index between the robot and the target is: (17) in, for The calculated value of the collision avoidance index between the robot and the moving target at any given time, which evaluates the minimum distance between any robot and the target; For any robot and any moving target in The minimum distance at any given moment; The desired separation distance between the robot and the target; This refers to the critical safety distance between the robot and the target; here, , .
[0052] The expression for the collision avoidance index between the robot and obstacles is: (18) in, Let be the calculated value of the collision avoidance index between the robot and the obstacle at time t. For any robot to reach any obstacle boundary in The shortest distance at any given moment is the minimum distance between any robot and an obstacle. The desired separation distance between the robot and the obstacle; This represents the minimum safe distance between the robot and the obstacle. Here, , The three minimum distances are defined. , and These minimum distances capture the worst-case approach scenarios and serve as a key indicator for collision risk assessment.
[0053] All safety metrics use a smooth transition function. Its definition is: (19) The goal is to find the set of parameters that maximizes the value of this fitness function. : (20) Specifically, the framework proposed in this application integrates a real-time parameter optimization mechanism that can continuously adjust the swarm control parameters according to environmental changes. This adaptive capability enables the robot swarm to dynamically optimize its behavior when facing diverse scenarios in tracking tasks. The entire multi-target tracking framework integrates a density-interaction-based swarm control model with a dynamic parameter optimization method, forming a complete adaptive control system, such as... Figure 3 As shown, it comprises four core components: external information, perception module, interactive behavior module, and Dynamic Particle Swarm Optimization (DPSO) algorithm.
[0054] The framework operates continuously in a closed-loop manner, starting with the external information acquisition phase—where the robot collects local information about neighboring robots, moving targets, and obstacles within its perception range. This is followed by the perception phase, where the system processes the acquired information. Considering the robot's limited field of vision and perception range, this phase directly impacts the quality and completeness of the data available to each individual robot. (During the perception phase, the system determines each robot's set of interactive neighbors, i.e....) , , Then, the interactive behavior phase begins, where the robot adjusts its individual behavior according to the aforementioned density-based interaction-based swarm control model. This density-based interaction-based swarm control model encompasses the interactions between robots, the interaction strategies between robots and targets, and the collision avoidance mechanisms between robots and obstacles.
[0055] This application framework employs a particle swarm optimization (PSO-FOHE) algorithm with fractional-order velocity and historical-guided estimation in each iteration. The algorithm dynamically optimizes the control parameters to determine the optimal set of control parameters that maximizes tracking performance. The optimization objective is defined by a fitness function: inputting a set of parameters into a density-interaction-based swarm control model outputs a corresponding fitness value. Therefore, the system's goal is to find a set of parameters using the PSO-FOHE algorithm. This maximizes the corresponding fitness value—the higher the value, the better the overall performance of cluster tracking.
[0056] After determining the optimal control parameters, the system calculates the control input for each robot and adjusts its speed and position in real time according to the interaction model. This process is executed cyclically at each time step, giving the system the ability to continuously adapt to changes in the environment. This integrated framework ensures the system has high adaptability and robustness in complex obstacle environments, maintaining an effective tracking configuration even when dealing with moving targets. By integrating local perception interaction mechanisms with efficient dynamic optimization strategies, the system achieves excellent swarm performance without relying on global information or centralized control, making it particularly suitable for multi-robot collaborative applications.
[0057] To effectively address the dynamic optimization problem of multi-object tracking, an enhanced particle swarm optimization algorithm (PSO-FOHE algorithm) is employed, featuring fractional-order velocity and history-guided estimation. This algorithm combines traditional PSO with fractional calculus principles to improve convergence in dynamic environments. When environmental changes are detected (indicated by significant changes in fitness evaluation), the algorithm activates the history-guided estimation mechanism, retrieving relevant solutions from the archive H of previously successful parameter configurations. This allows the swarm to quickly recover high-quality solutions when encountering similar patterns, significantly reducing computational overhead.
[0058] The computational complexity of DODM is ,in, It is the population size of PSO; It is the size of the PSO population; It optimizes the number of iterations; 、 and These represent the maximum number of neighboring robots, moving targets, and obstacles, respectively. The history-guided estimation mechanism significantly reduces the computational burden by 40-60% in repetitive scenarios, making the method suitable for real-time implementation.
[0059] Step S4: Substitute the optimized values of the control parameters at the current moment into the initial expression of the density-interaction-based swarm control model to obtain the objective expression of the density-interaction-based swarm control model at the current moment.
[0060] Step S5: Take any robot in the swarm as the current robot, and based on the local perception data of the current robot at the current moment and the target expression of the density-interaction-based swarm control model at the current moment, determine the estimated value of the control quantity of the current robot at the current moment, and then determine the estimated value of the control quantity of all robots at the current moment.
[0061] Specifically, after determining the interactive neighbors of each individual, the relevant parameters are substituted into the density-based group control model for calculation to obtain the control quantity of all robots at the current moment. The estimated value.
[0062] Step S6: Based on the estimated control values of all robots at the current moment, update the speed and position of all robots at the next moment to achieve obstacle avoidance while tracking multiple moving targets.
[0063] Specifically, calculation The acceleration is obtained by comparing the acceleration to the mass *m*. Multiplying this acceleration by the time interval yields the velocity increment. Then, adding the velocity increment to the velocity from the previous moment updates the velocity for the next moment, thus completing the velocity state update. Finally, the updated position is determined based on the updated velocity. The formulas for velocity and position updates are as follows: (twenty one) (twenty two) in, For robots speed; For robots exist The speed of time; For robots The velocity at time t-1; For time intervals; For robots exist The location at any given moment.
[0064] Furthermore, the convergence and stability of the swarm robot multi-target cooperative tracking method for highway inspection in this application are further demonstrated. Using Lyapunov stability theory, it is proved that: (1) the robots maintain safe distances from each other, targets, and obstacles for all ,have 、 and , (2) The system converges to a stable multi-target tracking configuration when the safe distance threshold is reached. hour, and These results validate the robustness of the swarm robot multi-target cooperative tracking method proposed in this application for highway inspection.
[0065] Beneficial effects: (1) By constructing a density-based group control model and introducing special action terms (i.e., various group terms) for obstacle avoidance and target approach, the system can automatically allocate robot resources among multiple targets without relying on explicit task allocation algorithms. This self-organizing mechanism enables the robot group to naturally form subgroups according to environmental conditions and target distribution, realizing dynamic optimization of resource allocation and significantly improving the system's adaptability, task execution efficiency and robustness.
[0066] (2) The density-based interaction mechanism not only covers simple pairwise interactions, but also comprehensively considers the spatial distribution characteristics of robots, enabling robots to effectively bypass obstacles during navigation while maintaining a safe distance from the target and other robots. This mechanism supports both tracking tasks and obstacle avoidance requirements in complex obstacle environments, achieving more adaptive group collaborative behavior.
[0067] (3) An enhanced particle swarm optimization algorithm based on fractional-order velocity and historical guidance estimation was adopted to construct a dynamic parameter optimization framework. This framework can continuously adjust the control parameters according to environmental changes, effectively balancing multiple competing objectives such as configuration quality, obstacle avoidance capability and target tracking efficiency. Moreover, the introduction of the historical guidance mechanism significantly reduces the computational overhead in repetitive scenarios, enabling the system to have the ability to respond in real time and continuously optimize.
[0068] (4) The convergence of the system was analyzed using Lyapunov stability theory, proving that the system can converge to the optimal tracking configuration while maintaining key safety constraints, and can remain stable even under dynamic environments and different noise levels. This theory supports the combination of self-organization and dynamic optimization mechanisms, which together ensure the robust performance of the system under different population sizes and disturbance conditions.
[0069] Based on the same inventive concept, this application also provides a swarm robot multi-target cooperative tracking system for highway inspection, which implements the aforementioned method for multi-target cooperative tracking of swarm robots for highway inspection. The solution provided by this system is similar to the implementation described in the above method. Therefore, the specific limitations of one or more embodiments of the swarm robot multi-target cooperative tracking system for highway inspection provided below can be found in the limitations of the swarm robot multi-target cooperative tracking method for highway inspection described above, and will not be repeated here.
[0070] In one exemplary embodiment, a swarm robot multi-target cooperative tracking system for highway inspection is provided, comprising: The data acquisition unit is used to acquire the local perception data of each robot in the swarm robot tracking system at the current moment. The local perception data includes: the position and velocity of all robots themselves, the position of all neighboring robots within the perception radius of each robot, the position of all moving targets, and the geometric information of all obstacles.
[0071] The initial model building unit is used to establish the initial expression of the density-interaction-based swarm control model; the density-interaction-based swarm control model is a function of control parameters, robot-to-robot interaction terms, robot-to-target interaction terms, robot-to-robot collision avoidance terms, target collision prevention terms, obstacle collision avoidance terms, and speed adjustment terms.
[0072] The control parameter optimization unit is used to dynamically optimize the control parameters in the density-interaction-based swarm control model using an enhanced particle swarm optimization algorithm, so as to obtain the optimized values of the control parameters at the current time.
[0073] The target model determination unit is used to substitute the optimized values of the control parameters at the current time into the initial expression of the density-interaction-based swarm control model to obtain the target expression of the density-interaction-based swarm control model at the current time.
[0074] The control quantity estimation unit is used to take any robot in the swarm as the current robot, and based on the local perception data of the current robot at the current moment and the objective expression of the density-interaction-based swarm control model at the current moment, determine the estimated value of the control quantity of the current robot at the current moment, and then determine the estimated value of the control quantity of all robots at the current moment.
[0075] The speed and position update unit is used to update the speed and position of all robots in the next moment based on the estimated control values of all robots at the current moment, so as to achieve obstacle avoidance while tracking multiple moving targets.
[0076] In one exemplary embodiment, a computer device is provided, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a swarm robot multi-target cooperative tracking method for highway inspection.
[0077] In one exemplary embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements a swarm robot multi-target cooperative tracking method for highway inspection.
[0078] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 4 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and databases. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network. When the computer program is executed by the processor, it implements a multi-target cooperative tracking method for swarm robots used in highway inspection.
[0079] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0080] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.
[0081] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).
[0082] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.
[0083] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0084] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods, systems, and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.
Claims
1. A method for multi-target cooperative tracking using swarm robots for highway inspection, characterized in that, The swarm robot multi-target cooperative tracking method for highway inspection includes: Acquire the local perception data of each robot in the swarm robot tracking system at the current moment; the local perception data includes: the position and velocity of all robots themselves, the position of all neighboring robots within the perception radius of each robot, the position of all moving targets, and the geometric information of all obstacles; Establish an initial expression for a density-interaction-based swarm control model; the density-interaction-based swarm control model is a function of control parameters, inter-robot interaction terms, robot-target interaction terms, inter-robot collision avoidance terms, target collision prevention terms, obstacle collision avoidance terms, and speed adjustment terms; An enhanced particle swarm optimization algorithm is used to dynamically optimize the control parameters in a density-interaction-based swarm control model, obtaining the optimized values of the control parameters at the current time step. Substituting the optimized values of the control parameters at the current moment into the initial expression of the density-interaction-based swarm control model, we obtain the objective expression of the density-interaction-based swarm control model at the current moment. Taking any robot in the swarm as the current robot, based on the local perception data of the current robot at the current moment and the objective expression of the swarm control model based on density interaction at the current moment, the estimated value of the control quantity of the current robot at the current moment is determined, and then the estimated value of the control quantity of all robots at the current moment is determined. Based on the estimated control values of all robots at the current moment, the velocity and position of all robots at the next moment are updated to enable tracking of multiple moving targets while avoiding obstacles.
2. The method according to claim 1, wherein, The initial expression for the density-interaction-based group control model is: ; in, For robots The estimated value of the control quantity; For robots Inter-robot interaction items; For robots Interaction items between the robot and the target; For robots Robot collision avoidance; For robots Target collision prevention items; For robots Obstacle avoidance items; For robots Speed adjustment option.
3. The multi-target cooperative tracking method for group robots for highway inspection according to claim 2, characterized in that, The control parameters include: inter-robot density interaction strength coefficient, robot-target density interaction strength coefficient, robot-robot collision avoidance repulsion strength coefficient, robot-target point repulsion strength coefficient, robot-obstacle repulsion strength coefficient, and reference density.
4. The multi-target cooperative tracking method for group robots for highway inspection according to claim 3, characterized in that, The expression for the robot-to-robot interaction term is: ; wherein, is the inter-robot density interaction strength coefficient; is the gradient operator of the robot local space density; is the reference density; is the gradient operator of the robot density; The expression for the interaction term between the robot and the target is: ; wherein, is the robot-target density interaction strength coefficient; is a cubic spline kernel function; k is the number of moving targets; is the robot set of moving targets within the perception radius; is the relative position vector of the k th moving target to the robot ; is the distance of the k th moving target to the robot ; is the perception radius; The expression for the robot collision avoidance term is: ; in, The number assigned to the neighboring robot; For robots A collection of neighboring robots within the perception radius; The robot-to-robot collision avoidance repulsion strength coefficient; For robots Neighbor Robot The relative position vectors between them; The expression for the target collision prevention term is: ; in, The robot-target point repulsion strength coefficient; The expression for the obstacle collision avoidance term is: ; in, For robots A set of obstacles within the perception radius; The robot-obstacle repulsion strength coefficient; The table shows the data from the robot. To the obstacle The shortest vector on the boundary; The expression for the speed adjustment term is: ; in, This is the acceleration coefficient; It is the damping factor; For robots The speed.
5. The method for multi-target cooperative tracking of swarm robots for highway inspection according to claim 4, characterized in that, An enhanced particle swarm optimization algorithm is used to dynamically optimize the control parameters in a density-interaction-based swarm control model, obtaining the optimized values of the control parameters at the current time step. Specifically, this includes: Construct a parameter vector to be optimized, consisting of the density interaction strength coefficient between robots, the density interaction strength coefficient between robots and targets, the collision avoidance and repulsion strength coefficient between robots and robots, the repulsion strength coefficient between robots and targets, the repulsion strength coefficient between robots and obstacles, and the reference density. An enhanced particle swarm optimization algorithm is used to optimize the parameter vector to be optimized in real time, with the goal of maximizing the overall performance index of the swarm robot tracking system. The comprehensive performance indicators include: position uniformity index, numerical distribution index, connectivity index, collision avoidance index between robots, collision avoidance index between robot and target, and collision avoidance index between robot and obstacle.
6. The method for multi-target cooperative tracking of swarm robots for highway inspection according to claim 5, characterized in that, The expression for the comprehensive performance index is: ; in, The comprehensive performance index function represents the overall performance index function. Time and system state are The parameter vector to be optimized is Overall performance score at the time; for The status of the swarm robot tracking system at any given moment; for The vector of parameters to be optimized at time step; for The calculated value of the positional uniformity index at any given time; for Calculated values of the numerical distribution index at any given time; for The calculated value of the connectivity index at any given time; for The calculated value of the collision avoidance index between robots at any given time; for The calculated value of the collision avoidance index between the robot and the target at any given time; for The calculated value of the collision avoidance index between the robot and obstacles at any given time.
7. The method for multi-target cooperative tracking of swarm robots for highway inspection according to claim 6, characterized in that, The expression for the location uniformity index is: ; in, for The calculated value of the positional uniformity index at any given time; The total number of moving targets; To move the target A collection of robots that interact with each other; With moving targets The number of robots that interact; For robots Location; For the first k The location of the moving target; The expression for the numerical distribution index is: ; in, for The calculated value of the numerical distribution index at time. This represents the expected number of robots for each moving target; It revolves around the moving target k The actual number of robots; N The number of robots; The expression for the connectivity index is: ; in, for The calculated value of the connectivity index at any given time; The number of robots that interact with at least one moving target; The expression for the collision avoidance index between robots is: ; in, for The calculated value of the collision avoidance index between robots at any given time; For smooth transition functions; For any pair of robots in The minimum distance at any given moment; It is the desired separation distance between the robots; It is the minimum safe distance between the robots; The expression for the collision avoidance index between the robot and the target is: ; in, for The calculated value of the collision avoidance index between the robot and the moving target at any given time; For any robot and any moving target in The minimum distance at any given moment; The desired separation distance between the robot and the target; The critical safety distance between the robot and the target; The expression for the collision avoidance index between the robot and obstacles is: ; in, for The calculated value of the collision avoidance index between the robot and obstacles at any given time; For any robot to reach any obstacle boundary in The shortest distance at any given moment; The desired separation distance between the robot and the obstacle; This is the minimum safe distance between the robot and the obstacle.
8. A swarm robot multi-target cooperative tracking system for highway inspection, characterized in that, The swarm robot multi-target cooperative tracking system for highway inspection is used to implement the swarm robot multi-target cooperative tracking method for highway inspection as described in any one of claims 1-7, wherein the swarm robot multi-target cooperative tracking system for highway inspection comprises: The data acquisition unit is used to acquire the local perception data of each robot in the swarm robot tracking system at the current moment. The local perception data includes: the position and velocity of all robots themselves, the position of all neighboring robots within the perception radius of each robot, the position of all moving targets, and the geometric information of all obstacles. The initial model building unit is used to establish the initial expression of the density-interaction-based swarm control model; the density-interaction-based swarm control model is a function of control parameters, robot-to-robot interaction terms, robot-to-target interaction terms, robot-to-robot collision avoidance terms, target collision prevention terms, obstacle collision avoidance terms, and speed adjustment terms. The control parameter optimization unit is used to dynamically optimize the control parameters in the density interaction-based swarm control model using an enhanced particle swarm optimization algorithm to obtain the optimized values of the control parameters at the current time. The target model determination unit is used to substitute the optimized values of the control parameters at the current time into the initial expression of the density-interaction-based swarm control model to obtain the target expression of the density-interaction-based swarm control model at the current time. The control quantity estimation unit is used to take any robot in the swarm as the current robot, and based on the local perception data of the current robot at the current time and the objective expression of the density-interaction-based swarm control model at the current time, determine the estimated value of the control quantity of the current robot at the current time, and then determine the estimated value of the control quantity of all robots at the current time. The speed and position update unit is used to update the speed and position of all robots in the next moment based on the estimated control values of all robots at the current moment, so as to achieve obstacle avoidance while tracking multiple moving targets.
9. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the swarm robot multi-target cooperative tracking method for highway inspection as described in any one of claims 1-7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the multi-target cooperative tracking method for highway inspection using swarm robots as described in any one of claims 1-7.
Citation Information
Patent Citations
Unmanned aerial vehicle cluster control method of jackdaw-imitating pairing interaction mechanism
CN112034882A
Interaction system of quadruped robot, planning control method and medium
CN119148751A
