A multi-UAV collaborative path planning method for multi-target tracking

Abstract

A multi-UAV cooperative path planning method for multi-target tracking is disclosed, relating to the field of UAV path planning technology. The advantages of this invention are: it decomposes the multi-UAV multi-target tracking problem into a target state estimation problem, a task allocation problem, and a motion planning problem, simplifying the solution; the real-time task allocation method employs a quadratic allocation approach to first allocate a reasonable initial target set to the UAV swarm, then uses an effective tracking strategy to select the current tracking target, while simultaneously avoiding tracking the same target through communication, enabling the UAV swarm to collaboratively execute multi-target tasks; and it utilizes an optimization-based method to generate smooth trajectories by considering optimization indicators such as smoothness, dynamic feasibility, and anti-interference, while ensuring the necessary safe distance exists between all objects of interest.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) path planning technology, and more specifically to a multi-UAV collaborative path planning method for multi-target tracking. Background Technology

[0002] Tracking animals and understanding their habits is crucial for protecting endangered species and conserving biodiversity. In the past, scientists have used various methods to track animal targets, such as vision-based sensors, GPS-based tags, and VHF-based tags. Despite all these emerging technologies, VHF-based tags remain widely used and play a vital role today. This is due to their small size, light weight, and long operating time, making them suitable for tracking most animals. However, traditional radio tracking methods require technicians to carry bulky radio receivers to acquire signals transmitted by animal radio tags, which is undoubtedly a time-consuming and labor-intensive task. Furthermore, traditional methods face significant difficulties in tracking animals that move rapidly in real-time in special environments (e.g., underwater or underground burrows). Drones, capable of carrying payloads such as radio receivers and antennas, offer a promising solution. Their deployability and high mobility offer the potential for automated target tracking, significantly reducing the time, effort, and cost associated with traditional tracking methods. In addition, advancements in software-defined radio (SDR) are crucial. These devices facilitate rapid reconfiguration and simultaneous reception of a wide range of radio frequencies. SDRs have achieved commercial viability, providing a cost-effective, durable, and lightweight solution. This makes them ideal for integration into drones, enabling the tracking of multiple animals simultaneously.

[0003] Autonomous unmanned aerial vehicle (UAV) systems for wildlife tracking have yielded some groundbreaking results. These systems use information such as the RSSI or angle of arrival (AOA) of the tag's radio signals to locate VHF radiotagged animals. In addition to improving positioning accuracy and reducing tracking errors, integrating sophisticated trajectory planning algorithms into these UAVs can enhance autonomy. Despite these technological advances, realizing UAVs capable of autonomously and effectively tracking wildlife remains a long-term task. Two unavoidable problems exist in achieving autonomous target tracking:

[0004] 1) Limited Onboard Resources: In practical applications, drones face challenges such as limited navigation battery life, limited onboard computing power, and limited payload capacity. Aerial robots should be able to fly out, track, and locate animals, and autonomously return to base without the need for battery replacements or human intervention. This highlights the importance of optimizing the operational efficiency of drones in wildlife tracking applications.

[0005] 2) Avoid Disturbance to Animals: The operation of drones in wildlife tracking can disturb animals, which is counterproductive to the mission objectives. Therefore, strategies must be designed not only to ensure the effectiveness of drones in tracking wildlife, but also to minimize any negative impact on the animals.

[0006] Unmanned aerial vehicle (UAV) planning methods can generally be categorized into information gain-based planning methods and task-based planning methods. Information gain-based planning methods use the POMDP framework to compute the optimal action to maximize reward. The reward function can be calculated using various methods, including task-driven and information-driven strategies. When uncertainty is high, information-driven strategies are more suitable for reducing target position uncertainty. Therefore, it is common to use information-driven strategies to compute the reward function. Several methods exist for measuring information divergence, such as Shannon entropy, Cauchy-Schwarz divergence, or Renyi divergence. However, information gain-based methods are computationally intensive and are generally unsuitable for real-time applications, / or require high-powered ground control systems for in-loop computation.

[0007] Task-based planning methods are lightweight, real-time approaches that can be embedded into low-cost vehicle-mounted UAV platforms. These methods demonstrate that the primary objective of tracking tasks is to effectively observe the target and minimize uncertainty in target state estimation. Therefore, prioritizing strategies that track targets with the lowest uncertainty in position estimation can provide inexpensive solutions. Task-based planning strategies often require the selection of control actions. It first tracks the target with the lowest uncertainty and, while maintaining a safe distance from the target, obtains a series of actions. Then, it evaluates each action using an evaluation function to obtain the optimal control action. In methods based on selecting control actions, they require advance planning of control actions and evaluation of all control actions using an evaluation function in each plan. However, all control actions may fail to meet the requirements, necessitating the reconsideration of new control actions. While evaluating control actions is computationally simple, designing control actions is more complex.

[0008] Currently, there is relatively more research on single UAVs tracking multiple moving targets, but research on multi-UAV cooperative tracking of multiple moving targets is limited. Multiple UAVs can effectively improve tracking efficiency and are therefore worthy of research. Summary of the Invention

[0009] The purpose of this invention is to address the shortcomings and defects of existing technologies by providing a multi-UAV cooperative path planning method for multi-target tracking. This method decomposes the multi-UAV multi-target tracking problem into a target state estimation problem, a task allocation problem, and a motion planning problem, thereby simplifying the solution to the multi-UAV multi-target tracking problem. The real-time task allocation method employs a quadratic allocation approach to first allocate a reasonable initial target set to the UAV swarm, and then uses an effective tracking strategy to select the current tracking target. Simultaneously, communication is used to avoid tracking the same target, enabling the UAV swarm to collaboratively execute multi-target tasks. Furthermore, an optimization-based method is used to generate a smooth trajectory by considering optimization indicators such as smoothness, dynamic feasibility, and anti-interference, while ensuring that the required safe distance exists between all objects of interest.

[0010] To achieve the above objectives, the present invention adopts the following technical solution: a multi-UAV cooperative path planning method for multi-target tracking, considering n r drone The system consists of components that jointly track n t One goal An algorithm is designed to facilitate efficient multi-UAV target tracking tasks for UAVs. A hierarchical framework is used to simplify the multi-target tracking problem in multi-UAV systems. The hierarchical framework consists of three layers: a state estimation layer, a real-time task allocation layer, and a motion planning layer. The state estimation layer utilizes simple and fast RSSI measurements and employs a particle filter algorithm to collect and estimate target positions. The real-time task allocation layer uses a quadratic allocation method to solve the multi-UAV task allocation problem in dynamic scenarios. The motion planning layer uses an optimization-based method to obtain a smooth trajectory that the UAV can execute, while ensuring that the UAV can maintain the expected safe distance from all objects of interest. This process mainly involves path planning and trajectory optimization. Initially, the enhanced RRT* algorithm considers the estimated position and safe distance of the target to calculate the initial UAV path points. Then, trajectory optimization considers optimization terms such as control force, dynamic feasibility, and interference avoidance to generate the final trajectory. The specific process is as follows: In the state estimation layer, the RSSI signal emitted by the UAV carrying the radio tag is first collected using a VHF antenna. Then, by using SDRs, the UAV... The drone can simultaneously and quickly configure radio signals from multiple targets and receive radio signals. The radio signals are processed by a signal detection algorithm. Since each target is uniquely identified by the radio transmission frequency, there is no data correlation problem. The target state estimation problem can be solved by running multiple particle filters in parallel. The task allocation layer adopts a quadratic allocation method to solve the multi-UAV task allocation problem in dynamic scenarios. First, the task pre-allocation problem is solved by a mixed integer optimization problem. Based on the known information of the UAVs and targets, the optimization problem is constructed by multiple constraints (such as distance constraints and UAV coordination constraints) to maximize the effectiveness of the task. Here, the initial position of the target can be obtained through multi-target region search. By solving this optimization problem, the initial target set is reasonably allocated to the UAV swarm. Subsequently, in order to more accurately locate and track moving targets, it is necessary to real-time reassign appropriate tracking targets to each UAV. The basic requirement of task reassignment is to avoid UAVs tracking the same target, which can effectively reduce the waste of UAV onboard resources and effectively complete the MCTT problem. After determining the position of the target to be tracked, the motion planning layer generates an executable trajectory for the UAV. An optimization-based method is adopted, with smoothness, dynamic feasibility, and interference avoidance as optimization terms. The trajectory is obtained by solving the optimization problem, thus obtaining Algorithm 1.

[0011] Furthermore, the particle filter specifically refers to a filtering problem where the objective is to filter based on historical observation data z from time t=1 to time t=k. 1:k To estimate the confidence level π k (x k ∣z 1:k Based on the Bayesian recursive approach, starting from the initial confidence level π0, the confidence density is calculated according to the order of prediction and update steps, as shown below: Particle filtering implements the random sampling process of the Monte Carlo method, using independent and uniformly distributed (iid) particles. The confidence level is approximated by a weighted set, i.e.: Where δ(·) represents the Dirac function, and the particle weights satisfy the condition By employing the position estimation uncertainty represented by the filter confidence level to determine the motion planning objective, although the E-optimal uncertainty estimate can be obtained using the maximum eigenvalue of the complete covariance matrix of the particles, a less computationally expensive method is used. The maximum covariance of the estimated position along each coordinate axis is calculated to represent the position uncertainty, exhibiting similar performance. It is a particle set The estimated object location, σ (x) Let π(·) represent the standard deviation of the estimated confidence density. Then:

[0012]

[0013] Furthermore, Algorithm 1 is designed based on this hierarchical framework, and its specific content is as follows:

[0014] MCTT problem algorithm;

[0015] Input: Initial position set of the UAV Target initial position set Target radio frequency set Initial target Stdσ0, final target Stdσ min Number of particles N, target motion model f(x) k |x k-1 ), observation model g(z) k |x k );

[0016] Output: UAV location set

[0017] (1) k=0,

[0018] (2) Through get

[0019] (3) ←Task pre-assignment

[0020] (4)

[0021] (5) k = k + 1;

[0022] (6) for each drone r ido;

[0023] (7) for each target do;

[0024] (8) Obtain the measured values

[0025] (9) ←State estimation

[0026] (10) Value Stored in variable Ω k ;

[0027] (11) ifσ tj <σ min then;

[0028] (12) The target t j from and Remove from;

[0029] (13) ←Task Reassignment

[0030] (14)

[0031] (15)

[0032] (16) Else;

[0033] (17) Drone i has completed all tasks and returned to the starting point.

[0034] Among them, the second row is based on the initial position set. and radio frequency sets of moving targets generate The drone swarm maintains communication in real time Update: When the drone completes tracking a specific target, it will... Delete the target when When empty, it indicates that all targets have been tracked; line (3) is based on and the initial position set of the drone Initial task set Pre-assigned to the drone swarm; line (8) obtains drone r at time k. i At target t j RSSI measurement value Based on this measurement The motion model of the target f(x) k |x k-1 ) and observation model g(z)k |x k Line (9) uses particle filtering on target t. j Perform state estimation to obtain the target t j Estimated location and confidence level (Expressed as the standard deviation of the estimated location (Std), see Section 4-b for details, then the value...) Stored in variable Ω k This is so that it can be used later in the redistribution algorithm when the target's estimated position is... Less than a certain value σ min When considering the localization and tracking of target tj in rows (11)-(12), based on the variable Ω that stores the estimated position of the target and Std at time k. k Total list Initial task set Initial Stdσ0 of the target and the list of targets tracked by the drone Line (13) performs task reassignment to obtain the drone r i Current targets and updates that need to be tracked In The executable trajectory of the target drone is generated in line (15) based on the location of the tracked target.

[0035] Furthermore, the real-time task allocation layer: First, based on the known initial positions of the UAVs and targets, considering conditions such as the minimum distance of the UAV swarm and the balance of targets to be executed, mixed integer programming is used to divide all targets in the area into UAVs. Then, each UAV obtains an initial target set. Next, the UAVs use a tracking strategy that prioritizes tracking the object with the lowest uncertainty in the estimated belief, where uncertainty can be represented by the filter confidence density. When the UAV is closer to the target, the signal strength received by the UAV from the target is greater, and the target position estimated using particle filtering is more accurate. This strategy achieves rapid detection and real-time tracking of radio-tagged targets. At the same time, due to their movement, the target may appear within the trackable range of other UAVs. In this case, it is necessary to avoid UAVs simultaneously tracking the same target through communication between UAVs. Specifically, the real-time task allocation layer is: Pre-allocation: This section establishes the multi-UAV task pre-allocation problem as an optimization problem to allocate reasonable initial tracking targets to UAVs, so that UAVs can work more effectively. Without losing generality, it is assumed that any task can be assigned to any UAV. Since each robot can execute at most Ni tasks, and one robot is needed to execute each task, all tasks to be executed should have The multi-drone task allocation problem is established as an optimization problem, aiming to achieve the optimal result with the minimum total cost while satisfying task constraints. This allows the drone swarm to better collaborate and improve work efficiency. Assume there are n... r One drone and n t Task objectives The drone's status is as follows: It is the three-dimensional coordinates of the drone. It is the yaw angle of the drone; the status of the mission target is... It uses Cartesian coordinates for the x, y, and z axes. Without loss of generality, it assumes that any task can be assigned to any drone, and that each robot can perform at most N tasks. i There are 10 tasks, and each task requires a single robot to perform. Therefore, for all the tasks to be performed, there should be 100 robots. The overall goal of task allocation is to assign all tasks to the robot to maximize total revenue or minimize total cost; let It is the allocation of task objectives t j Give drones r i The cost, since the order of task objectives is determined by the certainty of the estimated locations of the objectives, is not considered when allocating tasks; here, a ij Simplified to the distance between the drone and the mission target on the drone's flight plane, i.e. Therefore, the optimization objective is the cost required to execute all task objectives, expressed as: Where f ij To optimize the variable, let r represent i Is it assigned to t? j Therefore, the multi-robot task allocation problem can be formulated as an optimization problem: st, Here, constraint (1a) states that each task objective must be assigned to only one UAV and each task must be performed by one robot; constraint (1b) gives an upper limit to the number of task objectives that each UAV can perform, reflecting the limited nature of UAV resources; constraint (1c) states that f ij It is a binary variable representing r i Is it assigned to t? jReassignment: Although the task set has been assigned to the UAV, the target moves in real time. The UAV can identify other targets that it needs to track during its flight. At this time, dynamic task reassignment is triggered to make task execution more efficient. The principle of dynamic reassignment algorithm is identified as: moving to the target with the least uncertainty in state estimation. When the actual estimated position is unreliable, the UAV moving towards the target can quickly locate the target position. This is because when the UAV approaches the target, the signal strength increases and the signal is less affected by various interferences. Based on this principle, a dynamic task reassignment algorithm is proposed. The algorithm has high real-time performance and can meet the real-time task reassignment of UAV in dynamic environments.

[0036] Furthermore, the dynamic task reallocation algorithm is described as follows: For Each target t in j From Ω k Get its Std value It represents the uncertainty in the position estimate calculated by the previous particle filter algorithm, and then, if the Std of the target tj is less than σ0, it is stored in m. temp Unlike existing methods that select the estimated position with the lowest uncertainty among all targets, this method only considers the estimated position of the target when the target's Std is below a predetermined threshold. The basic principle is that the initial position estimates of particle filtering exhibit high uncertainty and significant changes over time. Using these estimates as the direct destination may lead to substantial trajectory changes and extended tracking duration. (The last sentence, "Delete m," appears to be an error and is left untranslated.) temp The target being tracked by other drones ( Except for the same target in the same category, to avoid the drone tracking the same target; if m temp If there is a target, select the target with the smallest Std for tracking; if m temp There was no target, and the drone was not yet on the pre-assigned target list. If the tracking target is completed, then in Select the nearest target to track; otherwise... An empty string means the drone has no objective to execute and can return to its starting point. Finally, now... Update

[0037] Furthermore, the motion planning layer specifically involves the following: When the actual position estimation is unreliable, the UAV navigating towards the target can quickly locate the target position because the uncertainty in target signal measurement is low when the UAV is close, as can be determined by distance tracking through signal strength measurement. Therefore, in tracking tasks targeting a specific object, a planning strategy that moves towards a target with minimal position estimation uncertainty may provide a cost-effective approach. After determining the target position based on the position estimation uncertainty, a trajectory needs to be generated. The basic requirements for the UAV's trajectory are smoothness, dynamic feasibility, and mutual avoidance between UAVs. By employing Σ... MINCO Trajectory generation is used to solve the above problems: the generation of the drone's trajectory is formulated as an unconstrained optimization problem. Where λ is the weight vector that weighs each cost function; control force J e : Control force optimization term J e It can be represented by integrating the norm squared term of the third derivative of the trajectory over time, written as... 2) Total time J t To improve the efficiency of drone swarm missions, it is desirable for the drone swarm to reach the target point as quickly as possible; therefore, the total time should be minimized. Obstacle Avoidance J o The Euclidean Symbolic Distance Field (ESDF) is used to establish the environment around the drone. ESDF can obtain the distance between the drone's current position and the nearest obstacle. To ensure safety, a collision penalty term is defined, which is triggered when the distance to the obstacle is less than the safety gap. The collision penalty term is as follows: Where, d thr It is a safety threshold. It considers the distance between a point and its nearest surrounding obstacle, and then obtains the obstacle avoidance optimization term J by calculating the weighted sum of the sampled constraint functions. o : in The orthogonal coefficients follow the trapezoidal rule; dynamic feasibility J d Because drones are constrained by their own motors and other actuators during flight, their speed and acceleration have maximum values. Therefore, a dynamic feasibility constraint term J is used. d Limit the maximum speed and acceleration to ensure the trajectory can be executed by the drone: J d =J v +J a #, Among them, v m ,a m These are the maximum permissible speed and acceleration; anti-interference J i: To maintain a safe distance r between the UAV and the target min , an anti-interference optimization term is constructed. Let V(u, r min ) denote the anti-interference region of the target state (position) based on the UAV state u, and it is given that: Then, the interruption avoidance penalty Ji is obtained by calculating the weighted sum of the sampling constraint functions as follows: The gradient of Ji with respect to ci and Ti is given by the following formula where t is the relative time on the trajectory. For the case where d(x, u) < dmin, the gradient is given by the following formula: where is the gradient of the Euclidean signed distance field (ESDF) in, otherwise, the gradient becomes

[0038] A multi-UAV cooperative path planning method for multi-target tracking, characterized in that it includes a target model, a UAV model and an observation model. Target model: For wildlife targets, their dynamic behaviors are usually unpredictable. Therefore, their behaviors are modeled as a random walk model where represents the state of the target at the k-th moment; represents a Gaussian distribution with mean μ and covariance Q; is the 3×3 covariance matrix of the process noise, where In represents the n×n-dimensional matrix; UAV model: For UAVs, the concept of multi-helicopter differential flatness has been widely explored and it has been proven that it has a significant and physically interpretable flat output space. All states of the UAV, including position, attitude, velocity and angular velocity, as well as inputs such as thrust acceleration and torque, can be represented by the flat output (Z) and its corresponding derivatives. It should be noted that the flat output space overlaps with the original configuration space, and the basic structure of its flat output is consistent with where (p x , p y , p z ) T represents the translation of the UAV's center of gravity, ψ is the yaw angle. By ensuring that the trajectory obeys a set of specific constraints in the flat space, it can be effectively transformed into the original configuration space using a planar transformation. Let represent the state of the UAV at the k-th moment, where are the 3D coordinates of the UAV, This refers to the drone's yaw angle. Here, the drone operates at a fixed altitude, focusing primarily on horizontal control. This method avoids adjusting the altitude during trajectory calculation, effectively reducing the computational load on the drone's onboard computer. Observation model: The observation model is a realistic signal propagation model used to estimate the target state x. k and drone status u k The RSSI measurement between the two points plays an important role in state estimation using a particle filter. The observation model uses a log-distance path loss model, and the power h(x) received by the UAV is... k ,u k (dBm) contains only the line-of-sight power component, h(x) k ,u k )=P0-10L c log(d(x k ,u k ))+G r (x k ,u k ), where P0(dBm) represents the reference power; L c This represents the unitless path loss constant, which characterizes signal attenuation as the transmission distance increases, typically ranging from 2 to 4. It is the distance between the target and the drone; G r (x k ,u k The yaw angle represents the gain of the directional antenna, which depends on the drone's yaw angle. and its relationship with target x k The relative position; the RSSI value zk (dBm) is easily corrupted by noise, including thermal noise and interference from other sources. Given the assumption of white noise, the measurement likelihood model can be expressed as: Among them, Q (z) This is the 1×1 covariance matrix for measuring noise.

[0039] After adopting the above technical solution, the beneficial effects of the present invention are as follows: The present invention has at least the following advantages:

[0040] 1) The multi-UAV multi-target tracking problem is decomposed into target state estimation problem, task allocation problem and motion planning problem to simplify the solution of multi-UAV multi-target tracking problem.

[0041] 2) The real-time task allocation method adopts a two-stage allocation method. First, a reasonable initial target set is allocated to the UAV swarm. Then, an effective tracking strategy is used to select the current tracking target. At the same time, communication is used to avoid tracking the same target, so that the UAV swarm can coordinately execute multi-target tasks.

[0042] 3) A trajectory generation method that utilizes optimization-based approaches to generate smooth trajectories by considering optimization metrics such as smoothness, dynamic feasibility, and anti-interference, while ensuring that the required safe distance exists between all objects of interest. Attached Figure Description

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

[0044] Figure 1 This is a layered framework diagram in this invention.

[0045] Figure 2 This is a schematic diagram of the positioning results of 10 moving targets in this invention.

[0046] Figure 3 yes Figure 2 Enlarged structural diagram at point A in the middle.

[0047] Figure 4 This is a schematic diagram of the time evolution Std value of each target tracked by the UAV in Scenario 1 of this invention.

[0048] Figure 5 This is a schematic diagram illustrating the changes in the RMS value of each tracked target in this utility model.

[0049] Figure 6 This is a diagram showing the results of four types of UAVs tracking and locating 10 moving targets in this invention.

[0050] Figure 7 This is a schematic diagram of the Std values ​​of the targets tracked by the four UAVs over time in this invention.

[0051] Figure 8 This is a schematic diagram illustrating the change of the RMS value of the target located by the four UAVs in this invention over time. Figure 1 .

[0052] Figure 9 This is a schematic diagram illustrating the change of the RMS value of the target located by the four UAVs in this invention over time. Figure 2 .

[0053] Figure 10 This is a schematic diagram illustrating the change of the RMS value of the target located by the four UAVs in this invention over time. Figure 3 .

[0054] Figure 11 This is a schematic diagram illustrating the change of the RMS value of the target located by the four UAVs in this invention over time. Figure 4 . Detailed Implementation

[0055] A multi-UAV cooperative path planning method for multi-target tracking, characterized in that it includes a target model, a UAV model, and an observation model:

[0056] Target model:

[0057] For wildlife targets, their dynamic behavior is often unpredictable; therefore, their behavior is modeled as a random walk model:

[0058]

[0059] in, This represents the state of the target at time k; This represents a Gaussian distribution with mean μ and covariance Q; It is the 3×3 covariance matrix of the process noise, where In represents an n×n dimensional matrix;

[0060] Drone Model:

[0061] For unmanned aerial vehicles (UAVs), the concept of multi-helicopter differential flatness has been extensively explored, and it has been proven to possess a significant, physically interpretable flat output space. All UAV states, including position, attitude, velocity, and angular velocity, as well as inputs such as thrust acceleration and torque, can be represented by a flat output (Z) and its corresponding derivatives. Notably, the planar output space overlaps with the original configuration space, and its basic structure of flat output is similar to... Consistent, where (p x ,p y ,p z ) T This represents the translation of the UAV's center of gravity, where ψ is the yaw angle. By ensuring that the trajectory adheres to a specific set of constraints in a flat space, it can be effectively transformed into the original configuration space using a planar transformation, allowing... Let $\mathbf{k}$ represent the state of the drone at time $k$. It is the 3D coordinates of the drone. This refers to the drone's yaw angle. Here, the drone uses a fixed flight altitude, focusing primarily on the drone's horizontal control. This method avoids adjusting the altitude during trajectory calculation, thus effectively reducing the computational load on the drone's onboard computer.

[0062] Observation model:

[0063] The observation model is a realistic signal propagation model used to estimate the target state x. k and drone status u kThe RSSI measurement between the two points plays an important role in state estimation using a particle filter. The observation model uses a log-distance path loss model, and the power h(x) received by the UAV is... k ,u k (dBm) includes only the line-of-sight power component:

[0064] h(x k ,u k )=P0-10L c log(d(x k ,u k ))+G r (x k ,u k ),

[0065] Where P0(dBm) represents the reference power; L c This represents the unitless path loss constant, which characterizes signal attenuation as the transmission distance increases, typically ranging from 2 to 4. It is the distance between the target and the drone; G r (x k ,u k The yaw angle represents the gain of the directional antenna, which depends on the drone's yaw angle. and its relationship with target x k The relative position;

[0066] The RSSI value zk (dBm) is easily corrupted by noise, including thermal noise and interference from other sources. Given the assumption of white noise, the measurement likelihood model can be expressed as:

[0067]

[0068] Among them, Q (z) This is the 1×1 covariance matrix for measuring noise.

[0069] Wild animals (or moving targets) are attached to VHF radio tags, each identified by a unique frequency. These radio collar tags transmit low-power on / off keying signals every T0 cycle, which are detected by a sensor system on the drone. Based on the strength of the received radio signals, the drone updates the position of the moving target using a particle filtering algorithm, and a motion planning algorithm generates an executable trajectory toward the target location.

[0070] Here, we introduce several assumptions to help define the research question in this paper:

[0071] (1) Since this paper focuses on the tracking and localization of the detected object rather than the signal detection problem, it assumes that there are no missed detections or false detections.

[0072] (2) This paper assumes that the UAV can detect all targets, which is reasonable in a medium-sized search area; this paper focuses on improving the tracking performance of detected targets.

[0073] (3) Assume that the total number of targets is known and that no target leaves the measurement area during the localization and tracking process. Therefore, the tracking and localization task terminates after all objects have been tracked and located.

[0074] (4) Assuming that the starting positions of all targets are known, these starting positions can be obtained through multi-UAV target search. This paper does not consider the exploration problem.

[0075] (5) In addition, in order to minimize disturbance to wildlife, the UAV must maintain a safe distance during navigation to track and locate wildlife. While UAVs can fly at high altitudes (e.g., above 100 meters) to minimize disturbance to objects of interest, it is also recommended to avoid flying over wildlife

[16] . Furthermore, flying at higher altitudes results in significantly shorter flight times, as it requires a certain amount of onboard battery power to lift the UAV to higher altitudes, especially for small UAVs. Therefore, in this paper, the flight altitude of the UAV is fixed at 30m to conserve the limited onboard battery power for tracking and search tasks, and changes in the UAV's altitude are not considered when calculating control actions, thereby reducing the computational requirements of the embedded computer.

[0076] Under the above assumptions, see Figure 1 As shown, the technical solution adopted in this specific embodiment is: considering n r drone The system consists of components that jointly track n t One goal Design an algorithm to facilitate UAVs to effectively perform multi-UAV target tracking tasks. Use a layered framework to simplify the multi-target tracking problem in multi-UAV systems. The layered framework can be divided into three layers: state estimation layer, real-time task allocation layer, and motion planning layer.

[0077] The state estimation layer utilizes simple and fast RSSI measurements and employs a particle filter algorithm to collect and estimate the target position;

[0078] The real-time task allocation layer uses a secondary allocation method to solve the problem of multi-UAV task allocation in dynamic scenarios;

[0079] The motion planning layer uses an optimization-based approach to obtain a smooth trajectory that the UAV can execute, while ensuring that the UAV can maintain the expected safe distance from all objects of interest. This process mainly involves path planning and trajectory optimization. Initially, the enhanced RRT* algorithm calculates the initial UAV waypoints by considering the estimated position and safe distance of the target. Then, trajectory optimization considers optimization terms such as control force, dynamic feasibility, and interference avoidance to generate the final trajectory.

[0080] The specific process is as follows:

[0081] In the state estimation layer, the RSSI signal carried by the radio tag is first acquired using a VHF antenna. Then, by using SDRs, the UAV can simultaneously and quickly configure radio and receive radio signals from multiple targets. The radio signals are processed by a signal detection algorithm. Since each target is uniquely identified by the frequency of radio transmission, there is no data correlation problem. The target state estimation problem can be solved by running multiple particle filters in parallel.

[0082] The task allocation layer employs a quadratic allocation method to address the multi-UAV task allocation problem in dynamic scenarios. First, the task pre-allocation problem is solved using a mixed integer optimization problem. Based on known information about the UAVs and targets, the optimization problem is constructed from multiple constraints (such as distance constraints and UAV coordination constraints) to maximize task effectiveness. Here, the initial position of the target can be obtained through multi-target region search. By solving this optimization problem, the initial target set is rationally allocated to the UAV swarm. Subsequently, to more accurately locate and track moving targets, it is necessary to reallocate appropriate tracking targets for each UAV in real time. The basic requirement for task reallocation is to avoid UAVs tracking the same target, which can effectively reduce the waste of UAV onboard resources and effectively complete the MCTT problem. We achieve this requirement by broadcasting the tasks executed by the UAVs.

[0083] After determining the location of the target to be tracked, the motion planning layer generates an executable trajectory for the UAV. An optimization-based approach is adopted, taking smoothness, dynamic feasibility and interference avoidance as optimization terms. The trajectory is obtained by solving the optimization problem, thus obtaining Algorithm 1.

[0084] As a more specific illustration of the present invention, the particle filter is specifically as follows:

[0085] The goal of the filtering problem is to filter based on historical observation data z from time t=1 to time t=k. 1:k To estimate the confidence level π k (x k ∣z 1:k Based on the Bayesian recursive approach, starting from the initial confidence level π0, the confidence density is calculated according to the order of prediction and update steps, as shown below:

[0086] π k∣k-1 (x k |z 1:k-1 )=

[0087] ∫f k∣k-1 (x k |x)π k-1 (x|z 1:k-1 )dx#

[0088]

[0089] Particle filtering implements the random sampling process of the Monte Carlo method, using independent and uniformly distributed (iid) particles. The confidence level is approximated by a weighted set, i.e.:

[0090]

[0091] Where δ(·) represents the Dirac function, and the particle weights satisfy the condition

[0092] By employing the position estimation uncertainty represented by the filter confidence level to determine the motion planning objective, although the E-optimal uncertainty estimate can be obtained using the maximum eigenvalue of the complete covariance matrix of the particles, a less computationally expensive method is used. The maximum covariance of the estimated position along each coordinate axis is calculated to represent the position uncertainty, exhibiting similar performance. It is a particle set The estimated object location, σ (x) Let π(·) represent the standard deviation of the estimated confidence density. Then:

[0093]

[0094]

[0095] As a more specific illustration of the present invention, Algorithm 1 is designed based on this hierarchical framework, and its specific content is as follows:

[0096] MCTT problem algorithm;

[0097] Input: Initial position set of the UAV Target initial position set Target radio frequency set Initial target Stdσ0, final target Stdσ min Number of particles N, target motion model f(x) k |x k-1 ), observation model g(z) k |x k);

[0098] Output: UAV location set

[0099] (1) k=0,

[0100] (2) Through get

[0101] (3) ←Task pre-assignment

[0102] (4)

[0103] (5) k = k + 1;

[0104] (6) for each drone r i do;

[0105] (7) for each target do;

[0106] (8) Obtain the measured values

[0107] (9) ←State estimation

[0108] (10) Value Stored in variable Ω k ;

[0109] (11)

[0110] (12) The target t j from and Remove from;

[0111] (13) ←Task Reassignment

[0112] (14)

[0113] (15)

[0114] (16) Else;

[0115] (17) Drone i has completed all tasks and returned to the starting point.

[0116] Among them, the second row is based on the initial position set. and radio frequency sets of moving targets generate The drone swarm maintains communication in real time Update: When the drone completes tracking a specific target, it will... Delete the target when When empty, it indicates that all targets have been tracked; line (3) is based on and the initial position set of the drone Initial task set Pre-assigned to the drone swarm; line (8) obtains drone r at time k. i At target t j RSSI measurement value Based on this measurement The motion model of the target f(x) k |x k-1 ) and observation model g(z) k |x k Line (9) uses particle filtering on target t. j Perform state estimation to obtain the target t j Estimated location and confidence level (Expressed as the standard deviation of the estimated location (Std), see Section 4-b for details, then the value...) Stored in variable Ω k This is so that it can be used later in the redistribution algorithm when the target's estimated position is... Less than a certain value σ min When considering the localization and tracking of target tj in rows (11)-(12), based on the variable Ω that stores the estimated position of the target and Std at time k. k Total list Initial task set Initial Stdσ0 of the target and the list of targets tracked by the drone Line (13) performs task reassignment to obtain the drone r i Current targets and updates that need to be tracked In The executable trajectory of the target drone is generated in line (15) based on the location of the tracked target.

[0117] As a more specific illustration of the present invention, the real-time task allocation layer: First, based on the known initial positions of the UAVs and targets, and considering conditions such as the minimum distance of the UAV swarm and the number of targets to be balanced, mixed integer programming is used to divide all targets in the area into UAVs. Then, each UAV obtains an initial target set. Next, the UAVs use a tracking strategy that prioritizes tracking the object with the lowest uncertainty in the estimated belief, where uncertainty can be represented by filter confidence density. When the UAV is closer to the target, the signal strength received by the UAV from the target is greater, and the target position estimated using particle filtering is more accurate. This strategy enables rapid detection and real-time tracking of radio-tagged targets. At the same time, due to its movement, the target may appear within the trackable range of other UAVs. In this case, communication between UAVs is needed to prevent UAVs from simultaneously tracking the same target. Specifically, the real-time task allocation layer is as follows:

[0118] 1) Pre-assignment: This section establishes the multi-UAV task pre-assignment problem as an optimization problem, assigning reasonable initial tracking targets to UAVs to enable them to work more efficiently. Without sacrificing generality, it is assumed that any task can be assigned to any UAV. Since each robot can execute at most Ni tasks, and one robot is needed to execute each task, all tasks to be executed should have...

[0119] By establishing the multi-drone task allocation problem as an optimization problem, we can obtain the optimization result with the minimum total cost under the condition of satisfying task constraints, so that the drone swarm can better cooperate and improve work efficiency.

[0120] Suppose there are n r One drone and n t Task objectives The drone's status is as follows: It is the three-dimensional coordinates of the drone. It is the yaw angle of the drone; the status of the mission target is... It uses Cartesian coordinates for the x, y, and z axes. Without loss of generality, it assumes that any task can be assigned to any drone, and that each robot can perform at most N tasks. i There are 10 tasks, and each task requires a single robot to perform. Therefore, for all the tasks to be performed, there should be 100 robots. The overall goal of task allocation is to assign all tasks to the robot to maximize total revenue or minimize total cost.

[0121] make It is the allocation of task objectives t j Give drones r iThe cost, since the order of task objectives is determined by the certainty of the estimated locations of the objectives, is not considered when allocating tasks; here, a ij Simplified to the distance between the drone and the mission target on the drone's flight plane, i.e. Therefore, the optimization objective is the cost required to execute all task objectives, expressed as: Where f ij To optimize the variable, let r represent i Is it assigned to t? j Therefore, the multi-robot task allocation problem can be formulated as an optimization problem:

[0122]

[0123] st

[0124]

[0125] Here, constraint (1a) states that each task objective must be assigned to only one UAV and each task must be performed by one robot; constraint (1b) gives an upper limit to the number of task objectives that each UAV can perform, reflecting the limited nature of UAV resources; constraint (1c) states that f ij It is a binary variable representing r i Is it assigned to t? j ;

[0126] Reassignment: Although the task set has been assigned to the UAV, the target moves in real time. The UAV can identify other targets that it needs to track during its flight. At this time, dynamic task reassignment is triggered to make task execution more efficient. The principle of dynamic reassignment algorithm is identified as: moving to the target with the least uncertainty in state estimation. When the actual estimated position is unreliable, the UAV moving towards the target can quickly locate the target position. This is because when the UAV approaches the target, the signal strength increases and the signal is less affected by various interferences. Based on this principle, a dynamic task reassignment algorithm is proposed. The algorithm has high real-time performance and can meet the real-time task reassignment of UAV in dynamic environments.

[0127] As a more specific illustration of the present invention, the dynamic task reallocation algorithm is described as follows:

[0128] Step 1: For Each target t in j From Ω k Get its Std value It represents the uncertainty in the position estimate calculated by the previous particle filter algorithm, and then, if the Std of the target tj is less than σ0, it is stored in m. tempIn this method, unlike existing methods that select the estimated position with the lowest uncertainty among all targets as the target, the estimated position of the target is only considered when the target Std is below a predetermined threshold. The basic principle is that the initial position estimation of particle filtering exhibits high uncertainty and significant changes over time. Using these estimates as the direct destination may lead to a large number of trajectory changes and extended tracking duration.

[0129] Step 2: Delete m temp The target being tracked by other drones ( (excluding) the same target in the same category, to avoid drones tracking the same target;

[0130] Step 3: If m temp If there is a target, select the target with the smallest Std for tracking; if m temp There was no target, and the drone was not yet on the pre-assigned target list. If the tracking target is completed, then in Select the nearest target to track; otherwise... An empty string means the drone has no objective to execute and can return to its starting point. Finally, now... Update

[0131] As a more specific illustration of the present invention, the motion planning layer specifically comprises:

[0132] In situations where actual position estimation is unreliable, a drone navigating to a target can quickly locate the target because when the drone is close, the uncertainty in the target signal measurement is low, which can be known from distance tracking achieved by measuring signal strength. Therefore, in tracking tasks with the target object as the objective, a planning strategy that moves towards the target with the lowest position estimation uncertainty may provide an inexpensive planning strategy.

[0133] After determining the target location to be tracked based on the uncertainty of the location estimation, a trajectory needs to be generated. The basic requirements for the trajectory of a UAV are smoothness, dynamic feasibility, and mutual avoidance between UAVs. In addition, there are additional requirements, such as minimizing execution time and minimizing interference with the target. By adopting Σ MINCO Trajectories can be used to solve the above problems:

[0134] The trajectory generation of the drone can be described as an unconstrained optimization problem:

[0135]

[0136] Where λ is the weight vector that weighs each cost function;

[0137] 1) Control force J e : Control force optimization term J e It can be represented by integrating the norm squared term of the third derivative of the trajectory over time, written as...

[0138]

[0139] 2) Total time J t To improve the efficiency of drone swarm missions, it is desirable for the drone swarm to reach the target point as quickly as possible; therefore, the total time should be minimized.

[0140]

[0141] 3) Obstacle Avoidance J o The Euclidean Symbolic Distance Field (ESDF) is used to establish the environment around the drone. ESDF can obtain the distance between the drone's current position and the nearest obstacle. To ensure safety, a collision penalty term is defined, which is triggered when the distance to the obstacle is less than the safety gap. The collision penalty term is as follows:

[0142]

[0143] Where, d thr It is a safety threshold. It considers the distance between a point and its nearest surrounding obstacle, and then obtains the obstacle avoidance optimization term J by calculating the weighted sum of the sampled constraint functions. o :

[0144]

[0145] in These are orthogonal coefficients that follow the trapezoidal rule;

[0146] 4) Dynamic feasibility J d Because drones are constrained by their own motors and other actuators during flight, their speed and acceleration have maximum values. Therefore, a dynamic feasibility constraint term J is used. d Limit the maximum speed and acceleration to ensure the trajectory can be executed by the drone:

[0147] J d =J v +J a #

[0148]

[0149]

[0150] Among them, v m ,a mis the maximum allowed speed and acceleration;

[0151] 5) Anti-interference J i : To maintain a safe distance r between the UAV and the target min , an anti-interference optimization term is constructed, letting V(u, r min ) represent the anti-interference area of the target state (position) based on the UAV state u, and given:

[0152]

[0153] Then, the interruption avoidance penalty Ji is obtained by calculating the weighted sum of the sampling constraint functions as

[0154]

[0155] The gradient of Ji with respect to ci and Ti is given by the following formula

[0156]

[0157] where t is the relative time on the trajectory. For the case of d(x, u) < dmin, the gradient is given by the following formula

[0158]

[0159] where is the gradient of the Euclidean signed distance field (ESDF) in the middle, otherwise, the gradient becomes

[0160] The following are the related embodiments of the present invention

[0161] In our simulation, 10 moving objects equipped with radio tags were set, randomly moving within a search area from 1000 meters to 1000 meters. The UAV always flew at a height of 30 meters above the ground, while the radio tags were placed at a height of 1 meter above the ground, thus restricting the planning to a two-dimensional domain. The maximum flight time was set to 3000 seconds. During the flight, a minimum safe distance of 50 meters was maintained between the UAV and the radio target to avoid interfering with the target. The particle filter used 10,000 particles. It should be noted that when the estimated uncertainty of the object was below the 20-meter threshold, the object was considered to be successfully tracked and located.

[0162] Here, we consider two scenarios. Scenario 1 describes a single drone located at [50,50,30] locating and tracking all targets, while Scenario 2 shows four drones located at [50,50,30], [1000,50,30], [50,1000,30], and [1000,1000,30]. The algorithm is evaluated based on the following metrics:

[0163] 1) Root Mean Square (RMS) (m): Estimates the distance between the target location and its actual location. Used... To represent the estimated location of the target, The actual position of the target is represented by RMS, which is the square root of the mean square error, given by the following formula:

[0164]

[0165] 2) Standard deviation of the estimation results (Std / Stdev): An indicator for assessing the uncertainty of the estimated location; see [link to documentation]. Perform detailed calculations.

[0166] 3) Flight time (s): The time it takes for the UAV to search for and locate all assigned radio-tagged targets.

[0167] 4) Planning time (s): The time that the UAV needs to allocate to the target being tracked, and the optimal tracking position is calculated.

[0168] See Figures 2-3 As shown, the localization results for 10 moving targets are displayed, with details such as RMS and flight time annotated along with each target location. In this scenario, the UAV successfully localized all 10 moving targets within 542 seconds, maintaining a maximum error distance of less than 25 meters, except for outlier target 1, whose root mean square distance was 36.6907 meters. After localizing the final target (target 7) within 542 seconds, the UAV received a command to return to its starting point. These results highlight the effectiveness of our algorithm in real-time search and localization of multiple targets, completing the task in approximately 10 minutes. Furthermore, it is noteworthy that the UAV's trajectory consistently maintained a safe distance from the radio-tagged target. Simulation results validate the effectiveness of our proposed trajectory planning algorithm in preventing the UAV from getting too close to the target while autonomously tracking it.

[0169] See Figure 4The diagram illustrates the temporal evolution of the Std value for each target tracked by the UAV in Scenario 1. The data represents the uncertainty variation of different radio tags. However, over time, the UAV can effectively reduce the Std value, representing uncertainty below a predetermined threshold of 20 meters. When the Std value of a tracked target falls below the predefined threshold, the target is considered successfully located, and the UAV can then track other targets. Notably, even when the UAV tracks other targets, the Std value of local targets remains stable, facilitating the simultaneous tracking and location of multiple targets.

[0170] See Figure 5 The figure shows the RMS value variation for each tracked target. The graph clearly demonstrates that at the time of localization, the RMS values ​​of most targets remained below 25 meters, with minimal changes observed thereafter. This validates the high accuracy of our tracking algorithm in locating all radio-tagged targets.

[0171] See Figure 6 The results shown illustrate the tracking and localization of 10 moving targets by four different UAVs. Similar to Scenario 1, estimation details are marked next to the target locations, including two metrics: RMS and time of flight. It can be seen that the UAV swarm localizes all 10 moving targets with a maximum error distance of less than 20 meters in 197 seconds. These results demonstrate that our algorithm can effectively collaborate with UAV swarms to solve the MCTT problem, significantly reducing the task execution time and achieving similar tracking performance to a single UAV tracking and localizing all targets.

[0172] See Figure 7 The figure illustrates the Std values ​​of targets tracked by four UAVs over time. The results show that the estimation uncertainty of the radio tags tracked by each UAV differs; however, a planning strategy can effectively reduce the estimation uncertainty to achieve radio tag localization.

[0173] See Figures 8-11 The data shows how the RMS values ​​of the targets located by the four drones change over time.

[0174] The results show that the RMS value can be reduced to the required value within a finite time, indicating that our algorithm enables UAVs to quickly locate designated radio marker targets.

[0175] The above description is only used to illustrate the technical solution of the present invention and is not intended to limit it. Any other modifications or equivalent substitutions made by those skilled in the art to the technical solution of the present invention, as long as they do not depart from the spirit and scope of the technical solution of the present invention, should be covered within the scope of the claims of the present invention.

Claims

1. A multi-UAV cooperative path planning method for multi-target tracking, characterized in that: Consider n r drone The system consists of components that jointly track n t One goal Design an algorithm to facilitate UAVs to effectively perform multi-UAV target tracking tasks. Use a layered framework to simplify the multi-target tracking problem in multi-UAV systems. The layered framework can be divided into three layers: state estimation layer, real-time task allocation layer, and motion planning layer. The state estimation layer utilizes simple and fast RSSI measurements and employs a particle filter algorithm to collect and estimate the target position; The real-time task allocation layer uses a secondary allocation method to solve the problem of multi-UAV task allocation in dynamic scenarios; The motion planning layer uses an optimization-based approach to obtain a smooth trajectory that the UAV can execute, while ensuring that the UAV can maintain the expected safe distance from all objects of interest. This process involves path planning and trajectory optimization. Initially, the enhanced RRT* algorithm calculates the initial UAV waypoints considering the estimated position and safe distance of the target. Then, trajectory optimization considers controllability, dynamic feasibility, and disturbance avoidance optimization terms to generate the final trajectory. The specific process is as follows: In the state estimation layer, the RSSI signal carried by the radio tag is first acquired using a VHF antenna. Then, by using SDRs, the UAV can simultaneously and quickly configure radio and receive radio signals from multiple targets. The radio signals are processed by a signal detection algorithm. Since each target is uniquely identified by the frequency of radio transmission, there is no data correlation problem. The target state estimation problem can be solved by running multiple particle filters in parallel. The task allocation layer employs a quadratic allocation method to address the multi-UAV task allocation problem in dynamic scenarios. First, the task pre-allocation problem is solved using a mixed integer optimization problem. Based on known information about the UAVs and targets, the optimization problem is constructed from multiple constraints, including distance constraints and UAV coordination constraints, aiming to maximize task effectiveness. The initial position of the target can be obtained through multi-target region search. By solving this optimization problem, the initial target set is reasonably allocated to the UAV swarm. Subsequently, in order to more accurately locate and track moving targets, it is necessary to reallocate appropriate tracking targets for each UAV in real time. The basic requirement of task reallocation is to avoid UAVs tracking the same target, which can effectively reduce the waste of UAV onboard resources and effectively complete the MCTT problem. After determining the location of the target to be tracked, the motion planning layer generates an executable trajectory for the UAV. An optimization-based approach is adopted, with smoothness, dynamic feasibility, and interference avoidance requirements as optimization terms. The trajectory is obtained by solving the optimization problem, thus yielding Algorithm 1, which is designed based on a hierarchical framework.

2. The multi-UAV cooperative path planning method for multi-target tracking according to claim 1, characterized in that: The particle filter is specifically: The goal of the filtering problem is to filter based on historical observation data z from time t=1 to time t=k. 1:k To estimate the confidence level π k (x k ∣z 1:k Based on the Bayesian recursive approach, starting from the initial confidence level π0, the confidence density is calculated according to the order of prediction and update steps, as shown below: Particle filtering implements the random sampling process of the Monte Carlo method, using independent and uniformly distributed particles. The confidence level is approximated by a weighted set, i.e.: Where δ(·) represents the Dirac function, and the particle weights satisfy the condition By employing the position estimation uncertainty represented by the filter confidence level to determine the motion planning objective, although the E-optimal uncertainty estimate can be obtained using the maximum eigenvalue of the complete covariance matrix of the particles, a less computationally expensive method is used. The maximum covariance of the estimated position along each coordinate axis is calculated to represent the position uncertainty, exhibiting similar performance. It is a particle set The estimated object location, σ (x) Let π(·) represent the standard deviation of the estimated confidence density. Then:

3. The multi-UAV cooperative path planning method for multi-target tracking according to claim 1, characterized in that: Algorithm 1 is designed based on this hierarchical framework, and its specific content is as follows: MCTT problem algorithm; Input: Initial position set of the UAV Target initial position set Target radio frequency set Initial target Stdσ0, final target Stdσ min Number of particles N, target motion model f(x) k |x k-1 ), observation model g(z) k |x k ); Output: UAV location set (1) (2) Through get (3) Task pre-assignment (4) (5) k = k + 1; (6) for each drone r i do; (7) for each target (8) Obtain the measured values (9) ←State estimation (10) Value Stored in variable Ω k ; (11) (12) The target t j from and Remove from; (13) ←Task Reassignment (14) (15) (16) Else; (17) Drone i has completed all tasks and returned to the starting point. Among them, the second row is based on the initial position set. and radio frequency sets of moving targets generate The drone swarm maintains communication in real time Update: When the drone completes tracking a specific target, it will... Delete the target when When empty, it indicates that all targets have been tracked; line (3) is based on and the initial position set of the drone Initial task set Pre-assigned to the drone swarm; line (8) obtains drone r at time k. i At target t j RSSI measurement value Based on this measurement The motion model of the target f(x) k |x k-1 ) and observation model g(z) k |x k Line (9) uses particle filtering on target t. j Perform state estimation to obtain the target t j Estimated location and confidence level Expressed as the standard deviation of the estimated location, then the value Stored in variable Ω k This is so that it can be used later in the redistribution algorithm when the target's estimated location is... Less than a certain value σ min When considering the localization and tracking of target tj in rows (11)-(12), based on the variable Ω that stores the estimated position of the target and Std at time k. k Total list Initial task set Initial Stdσ0 of the target and the list of targets tracked by the drone Line (13) performs task reassignment to obtain the drone r i Current targets and updates that need to be tracked In The executable trajectory of the target drone is generated in line (15) based on the location of the tracked target.

4. The multi-UAV cooperative path planning method for multi-target tracking according to claim 1, characterized in that: The real-time task allocation layer: First, based on the known initial positions of the drones and the targets, considering the minimum distance of the drone swarm and the condition of the balance number of targets to be executed, all the targets in the area are divided into drones using mixed integer programming. Then, each drone obtains an initial target set. Next, the drones use a tracking strategy that gives priority to tracking the object with the lowest estimated belief uncertainty, where the uncertainty can be represented by the filter confidence density. When the drone is closer to the target, the drone receives a stronger signal intensity emitted by the target, and the target position estimated using particle filtering is more accurate. This strategy achieves the rapid detection and real-time tracking of radio-tagged targets. At the same time, due to its movement, the target may appear within the trackable range of other drones. At this time, communication between drones is required to avoid drones tracking the same target simultaneously. Its real-time task allocation layer is specifically as follows: 1) Pre-assignment: The multi-UAV task pre-assignment problem is established as an optimization problem. It involves assigning reasonable initial tracking targets to the UAVs to enable them to work more efficiently. Without sacrificing generality, it is assumed that any task can be assigned to any UAV. Since each robot can execute at most Ni tasks, and one robot is needed to execute each task, all tasks to be executed should have... The multi-drone task allocation problem is formulated as an optimization problem, enabling an optimization result with the minimum total cost to be obtained under the premise of meeting the task constraints, so that the drone swarm can better perform teamwork and improve work efficiency; Suppose there are n r There are R drones = {r1, ..., r} i ,…,r nr } and n t Task objectives The drone's status is as follows: It is the three-dimensional coordinates of the drone. It is the yaw angle of the drone; the status of the mission target is... It uses Cartesian coordinates for the x, y, and z axes. Without loss of generality, it assumes that any task can be assigned to any drone, and that each robot can perform at most N tasks. i There are 10 tasks, and each task requires a single robot to perform. Therefore, for all the tasks to be performed, there should be 100 robots. The overall goal of task allocation is to assign all tasks to the robot to maximize total revenue or minimize total cost. make It is the allocation of task objectives t j Give drones r i The cost, since the order of task objectives is determined by the certainty of the estimated locations of the objectives, is not considered when allocating tasks; here, a ij Simplified to the distance between the drone and the mission target on the drone's flight plane, i.e. Therefore, the optimization objective is the cost required to execute all task objectives, expressed as: Where f ij To optimize the variable, let r represent i Is it assigned to t? j Therefore, the multi-robot task allocation problem can be formulated as an optimization problem: Here, constraint (1a) states that each task objective must be assigned to only one UAV and each task must be performed by one robot; constraint (1b) gives an upper limit to the number of task objectives that each UAV can perform, reflecting the limited nature of UAV resources; constraint (1c) states that f ij It is a binary variable representing r i Is it assigned to t? j ; 2) Reallocation: Although the task set has been assigned to the drones, the targets move in real time, and the drones can identify other targets that need to be tracked during their flight. At this time, dynamic task reallocation is triggered to make the task execution more efficient. By identifying the principle of the dynamic reallocation algorithm as: moving to the target with the least state estimation uncertainty. When the actual estimated position is unreliable, the drone moving towards the target can quickly locate the target position because when the drone approaches the target, the signal intensity increases and the signal is less affected by various disturbances. Based on this principle, a dynamic task reallocation algorithm is proposed, which has high real-time performance and can meet the real-time task reallocation of drones in a dynamic environment.

5. A multi-UAV cooperative path planning method for multi-target tracking according to claim 1, characterized in that: The described dynamic task reallocation algorithm is described as follows: S1: For Each target t in j From Ω k Get its Std value It represents the uncertainty in the position estimate calculated by the previous particle filter algorithm, and then, if the Std of the target tj is less than σ0, it is stored in m. temp In this method, unlike existing methods that select the estimated position with the lowest uncertainty among all targets as the target, the estimated position of the target is only considered when the target Std is below a predetermined threshold. The basic principle is that the initial position estimation of particle filtering exhibits high uncertainty and significant changes over time. Using these estimates as the direct destination may lead to a large number of trajectory changes and extended tracking duration. S2: Delete m temp The target being tracked by other drones To avoid drones tracking the same target, the same target must be excluded. S3: If m temp If there is a target, select the target with the smallest Std for tracking; if m temp There was no target, and the drone was not yet on the pre-assigned target list. If the tracking target is completed, then in Select the nearest target to track; otherwise... An empty string means the drone has no objective to execute and can return to its starting point. Finally, now... Update 6. A multi-UAV cooperative path planning method for multi-target tracking according to claim 1, characterized in that: The specific motion planning layer is as follows: In the case where the actual position estimation is unreliable, the drone navigating towards the target can quickly locate the target position because when the drone is closer, the uncertainty of the target signal measurement is lower, which can be known from distance tracking by measuring the signal intensity. Therefore, in the tracking task with the measured object as the target, aiming to achieve a sufficiently small position estimation uncertainty, the planning strategy of moving towards the target with the lowest position estimation uncertainty may provide a cheap planning strategy; After determining the target location to be tracked based on the uncertainty of the location estimation, a trajectory needs to be generated. The basic requirements for the trajectory of the UAV are smoothness, dynamic feasibility, and mutual avoidance between UAVs. This is achieved by adopting Σ MINCO The trajectory is used to solve the above problem: The trajectory generation of the drone is expressed as an unconstrained optimization problem: where λ is the weight vector that weighs each cost function; 1) Control force J e : Control force optimization term J e It can be represented by integrating the norm squared term of the third derivative of the trajectory over time, written as... 2) Total time J t To improve the efficiency of drone swarm missions, it is desirable for the drone swarm to reach the target point as quickly as possible; therefore, the total time should be minimized. 3) Obstacle Avoidance J o The Euclidean Symbolic Distance Field (ESDF) is used to establish the environment around the drone. ESDF can obtain the distance from the drone's current position to the nearest obstacle. To ensure safety, a collision penalty term is defined, which is triggered when the distance to the obstacle is less than the safety gap. The collision penalty term is as follows: Where, d thr It is a safety threshold. It considers the distance between a point and its nearest surrounding obstacle; Then, the obstacle avoidance optimization term J is obtained by calculating the weighted sum of the sampling constraint functions. o : in These are orthogonal coefficients that follow the trapezoidal rule; 4) Dynamic feasibility J d Because drones are constrained by their own motor actuators during flight, their speed and acceleration have maximum values. Therefore, a dynamic feasibility constraint term J is used. d Limit the maximum speed and acceleration to ensure the trajectory can be executed by the drone: J d *J v +J a Among them, v m ,a m These are the maximum permissible speed and acceleration; 5) Anti-interference J i To maintain a safe distance between the drone and the target min An anti-interference optimization term was constructed, letting V(u,r) min The region representing the anti-jamming zone based on the target state u of the UAV, including its position, is given as follows: Then, the interruption avoidance penalty Ji is obtained by calculating the weighted sum of the sampled constraint functions as The gradient of Ji with respect to ci and Ti is given by the following formula where t is the relative time on the trajectory. For the case of d(x,u) < dmin, the gradient is given by the following formula in, yes The gradient of the Euclidean symbolic distance field ESDF; otherwise, the gradient becomes 7. A multi-UAV cooperative path planning method for multi-target tracking, characterized in that, It includes a target model, a drone model, and an observation model Target model: For wild animal targets, their dynamic behaviors are usually unpredictable. Therefore, their behaviors are modeled as a random walk model in, This represents the state of the target at time k; This represents a Gaussian distribution with mean μ and covariance Q; It is the 3×3 covariance matrix of the process noise, where In represents an n×n dimensional matrix; Drone model: For unmanned aerial vehicles (UAVs), the concept of multi-helicopter differential flatness has been extensively explored, and it has been proven to possess a significant, physically interpretable flat output space. All UAV states, including position, attitude, velocity, and angular velocity, as well as thrust acceleration and torque input, can be represented by a flat output Z and its corresponding derivative. Notably, the planar output space overlaps with the original configuration space, and its basic structure of flat output is similar to... Consistent, where (p x ,p y ,p z ) T This represents the translation of the UAV's center of gravity, where ψ is the yaw angle. By ensuring that the trajectory adheres to a specific set of constraints in a flat space, it can be effectively transformed into the original configuration space using a planar transformation, allowing... Let $\mathbf{k}$ represent the state of the drone at time $k$. It is the 3D coordinates of the drone. This refers to the drone's yaw angle. Here, the drone uses a fixed flight altitude, focusing primarily on the drone's horizontal control. This method avoids adjusting the altitude during trajectory calculation, thus effectively reducing the computational load on the drone's onboard computer. Observation model: The observation model is a realistic signal propagation model used to estimate the target state x. k and drone status u k The RSSI measurement between the two points plays an important role in state estimation using a particle filter. The observation model uses a log-distance path loss model, and the power h(x) received by the UAV is... k ,u k Unit: dBm (includes only line-of-sight power component) h(x k ,u k )=P0-10L c log(d(x k ,u k ))+G r (x k ,u k ), Where P0 is in dBm, representing the reference power; L c This represents the unitless path loss constant, which characterizes signal attenuation as the transmission distance increases, typically ranging from 2 to 4. It is the distance between the target and the drone; G r (x k ,u k The yaw angle represents the gain of the directional antenna, which depends on the drone's yaw angle. and its relationship with target x k The relative position; RSSI value zk (unit: dBm) is susceptible to noise degradation, including thermal noise and interference from other sources. Given the assumption of white noise, the measurement likelihood model can be expressed as: Among them, Q (z) This is the 1×1 covariance matrix for measuring noise.

Images