Unmanned aerial vehicle bionic group trajectory emergence method and system of space-time gradient field

Abstract

The invention provides an unmanned aerial vehicle bionic group trajectory emergence method and system of a space-time gradient field, and relates to the technical field of unmanned aerial vehicle group collaborative navigation and adaptive control, and the method comprises the steps: generating a four-dimensional space-time gradient field tensor matrix through a dynamic field coupling convolutional network; in combination with density-speed response data of fish school escape behaviors and unmanned aerial vehicle dynamics constraints, a bionic group interaction rule set is constructed through nonlinear differential game modeling; adopting a space-time finite element discretization method and combining with GPU acceleration parallel solving to determine an individual trajectory basis function of the unmanned aerial vehicle; performing optimization processing through a Lyapunov function to generate a group recombination instruction; and generating a space-time gradient field enabled unmanned aerial vehicle bionic group trajectory emergence scheme by integrating the updated space-time gradient field, the group recombination instruction and the historical trajectory data. According to the invention, high-precision, low-delay and high-robustness autonomous navigation capability is provided for the unmanned aerial vehicle group, and the task success rate is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) swarm cooperative navigation and adaptive control. Specifically, it relates to a method and system for the emergence of UAV bionic swarm trajectories in a spatio-temporal gradient field. Background Art

[0002] Currently, UAV swarm cooperative navigation technology in complex environments mainly relies on centralized path planning algorithms and traditional bionic swarm intelligence models (such as Boids rules, particle swarm optimization). These technologies perform well in static or low-dynamic environments, but expose significant defects in complex dynamic scenarios: 1. Traditional methods rely on static or low-frequency updated obstacle maps and cannot respond in real time to dynamic expansions (such as the dynamic changes of ice cracks) and sudden interferences (instantaneous wind speed > 15 m / s). For example, the replanning delay of the RRT algorithm in the scenario of crack expansion is as high as 2 - 5 seconds, resulting in a significant increase in the risk of crashing. Moreover, existing bionic models lack a physical field coupling mechanism and cannot dynamically correct trajectories through environmental forces (such as wind force, ocean current), resulting in a disorder rate of more than 65% in the swarm under sudden wind disturbances; 2. Traditional multi-objective optimization (such as NSGA-II) does not consider the decrease in motor efficiency caused by complex environments, and the fixed weight allocation makes it difficult to balance energy consumption and obstacle avoidance. For example, in a sea surface or ice surface environment, low temperature or high humidity conditions will affect the motor performance and battery life of UAVs; 3. Existing GPU acceleration schemes (such as CUDA uniform grid) do not target risk grading. Single-precision calculation in the entire area results in a trajectory error > 0.5 m in high-risk areas, while over-computation occurs in safe areas. In addition, in a sea surface or ice surface environment, UAVs need to cope with dynamically changing ocean currents and wind fields, and existing technologies cannot effectively adapt to these dynamic environments. For example, when the environment suddenly changes, full-path replanning is required, and the time-consuming for a swarm of thousands of UAVs is > 10 minutes, seriously lagging behind the task requirements; 4. In an ice surface or sea surface environment, UAV swarm cooperative navigation also faces special challenges. For example, sea surface waves and ocean currents have a greater impact on the flight stability of UAVs, and existing navigation technologies lack effective modeling and compensation for these factors. Summary of the Invention

[0003] The purpose of the present invention is to provide a method and system for the emergence of UAV bionic swarm trajectories in a spatio-temporal gradient field to improve the above problems. To achieve the above purpose, the technical solutions adopted by the present invention are as follows: In the first aspect, the present application provides a method for the emergence of UAV bionic swarm trajectories in a spatio-temporal gradient field, including: Obtaining satellite remote sensing ice crack distribution data, LiDAR real-time terrain scanning data, ocean current velocity field data, surface temperature gradient data, and instantaneous wind speed vector data, and generating a four-dimensional spatio-temporal gradient field tensor matrix through a dynamic field coupling convolutional network; Based on the four-dimensional space-time gradient field tensor matrix, combined with the density-velocity response data of fish swarm escape behavior and the dynamic constraints of unmanned aerial vehicles (UAVs), a set of bionic group interaction rules is constructed through non-linear differential game modeling; Using the bionic group interaction rule set and the four-dimensional space-time gradient field tensor matrix, the spatio-temporal finite element discretization method is adopted and combined with GPU-accelerated parallel solution to determine the basis function of the UAV individual trajectory; According to the basis function of the UAV individual trajectory and the real-time updated data of the environmental sensor, through the optimization process of the Lyapunov function, the four-dimensional space-time gradient field tensor matrix is dynamically adjusted to obtain the updated space-time gradient field and generate a group reorganization instruction; Integrating the updated space-time gradient field, the group reorganization instruction and the historical trajectory data, applying the quantum annealing-NSGAIII hybrid optimization algorithm, a UAV bionic group trajectory emergence scheme empowered by the space-time gradient field is generated.

[0004] Preferably, the satellite remote sensing ice crack distribution data, LiDAR real-time terrain scanning data, ocean current velocity field data, surface temperature gradient data and instantaneous wind speed vector data are obtained, and a four-dimensional space-time gradient field tensor matrix is generated through a dynamic field coupling convolutional network, which includes: The ice surface image is obtained by the synthetic aperture radar carried by the polar orbiting satellite, and the binary map of the crack distribution with a spatial resolution ≤ 1m is extracted by the ice crack edge detection algorithm to obtain the satellite remote sensing ice crack distribution data; the original point cloud data is collected by the solid-state LiDAR carried by the UAV, and multiple frames of point clouds are registered by the point cloud registration algorithm to generate a digital elevation model to obtain the LiDAR real-time terrain scanning data; based on the Doppler current meters built in the terrain sensor array with a spacing of 500m, the horizontal velocity vector is measured in real time, and the ocean current velocity field data is generated by Kriging spatial interpolation; the ice surface temperature distribution matrix is obtained by the high-resolution infrared thermal imager carried by the UAV at a sampling rate of 5Hz, and the spatial gradient field is calculated based on the Sobel operator to obtain the surface temperature gradient data; the instantaneous wind speed vector data is obtained by Kriging interpolation calculation of the three-dimensional wind speed obtained by the complex environmental weather station network; The satellite remote sensing ice crack distribution data, LiDAR real-time terrain scanning data, ocean current velocity field data, surface temperature gradient data and instantaneous wind speed vector data are encoded into a five-channel spatio-temporal tensor, and the five-channel spatio-temporal tensor is used as the input of the dynamic convolutional network, so as to construct a dynamic convolutional neural network, the convolution kernel weights of which are jointly modulated by the ice crack curvature and the temperature gradient, the output is the coupled drift field vector, and combined with the coordinates and priority weights provided by the task requirement system, a four-dimensional space-time gradient field tensor matrix is generated, where the four-dimensional space-time gradient field tensor matrix includes a gravitational field, a repulsive field and a drift field.

[0005] Preferably, based on the four-dimensional space-time gradient field tensor matrix, combined with the density-velocity response data of the fish swarm escape behavior and the UAV dynamics constraints, a bionic swarm interaction rule set is constructed through non-linear differential game modeling, including: Based on the density-velocity observation data of the fish swarm escape behavior, the Sigmoid function is used to fit the non-linear mapping relationship between the density and the velocity, and a bionic density-velocity response constraint curve is generated; The bionic density-velocity response constraint curve and the preset UAV dynamics model are used to construct the Hamiltonian function, and a dynamically feasible solution space of the non-linear differential game is generated, where the UAV dynamics model is obtained by integrating the measured polar parameters, gradient field coupling and bionic rule constraints; Based on the ice crack repulsive force gradient field component in the four-dimensional space-time gradient field tensor matrix, the real-time relative positions of the UAV swarm are fused, and a combined force vector field for group obstacle avoidance and collision avoidance is generated through the inverse proportional repulsive potential superposition algorithm; According to the data of the dynamically feasible solution space, the combined force vector field for group obstacle avoidance and collision avoidance, and the drift field, through distributed Nash equilibrium solution and heading coordination optimization, a bionic swarm interaction rule set is obtained, including static constraints and dynamic coordination strategies.

[0006] Preferably, based on the ice crack repulsive force gradient field component in the four-dimensional space-time gradient field tensor matrix, the real-time relative positions of the UAV swarm are fused, and a combined force vector field for group obstacle avoidance and collision avoidance is generated through the inverse proportional repulsive potential superposition algorithm, including: Based on the ice crack repulsive force gradient field component in the four-dimensional space-time gradient field tensor matrix, the ice crack repulsive force gradient field covering the global mission area is calculated; The local repulsive force gradient in the ice crack repulsive force gradient field is extracted, and combined with the real-time position of the UAV, the anti-collision force between groups is generated through inverse proportional potential superposition; The vectors of the local repulsive force gradient and the anti-collision force between groups are superimposed to obtain the total obstacle avoidance and collision avoidance combined force of the UAV individuals.

[0007] Preferably, using the bionic swarm interaction rule set and the four-dimensional space-time gradient field tensor matrix, the space-time finite element discretization method and GPU-accelerated parallel solution are used to determine the trajectory basis function of the UAV individuals, including: According to the four-dimensional space-time gradient field tensor matrix and the bionic swarm heading coordination constraint, the Fourier-Legendre joint basis function expansion method is used to construct a space-time threat perception basis function set, the spatial wave number component of which is modulated by the ice crack distribution density, the time frequency component is correlated with the time-varying rate of the drift field, and the heading coordination angle is embedded in the basis function phase through the cosine modulation term; The space-time threat perception basis function set and the UAV dynamics model are discretized by the Galerkin weighted residual method to generate a trajectory coupling algebraic system , where the coefficient matrix is composed of the inner products of basis functions, and the right-hand side includes the orthogonal projection of the group collision avoidance force; Based on the trajectory coupling algebraic system and the four-dimensional spatio-temporal gradient field tensor matrix, implement the GPU parallel solution processing for ice gap risk grading to obtain the risk zoning strategy. The risk zoning strategy includes a high-risk area within 0.5 m of the ice crack edge, where a 32×32 thread block is allocated, a medium-risk area in the 1-2 m area outside the crack, where a 64×64 thread block is allocated, and a flat ice surface area, where a 128×128 thread block is allocated. In the high-risk area, double-precision floating-point operations and preconditioning acceleration are used, and in the medium- and low-risk areas, single-precision hybrid iterative calculations are used; Output the basis function coefficient vector through the risk zoning strategy, thereby generating the continuous trajectory function of the unmanned aerial vehicle.

[0008] Specifically, the comprehensively updated spatio-temporal gradient field, group reorganization instructions, and historical trajectory data are applied to the quantum annealing-NSGAIII hybrid optimization algorithm to generate a spatio-temporal gradient field-enabled emergence scheme for the bionic swarm trajectory of unmanned aerial vehicles, including: Based on the updated spatio-temporal gradient field, group reorganization instructions, and historical trajectory data, construct a multi-objective problem that includes minimizing energy consumption, minimizing the number of ice crack collisions, and optimizing task synchronization. Among them, the control input is restricted by the maximum thrust of the unmanned aerial vehicle and the bionic speed constraint. The ice crack collision determination is triggered by the curvature field threshold, and the task synchronization is measured by the range of arrival times; Use the quantum annealing mechanism to generate the initial population. The Hamiltonian design couples the historical trajectory conflict statistics and the alignment degree of the repulsive force field gradient direction. Each individual encodes the control input sequence and heading strategy of the unmanned aerial vehicle. The quantum bit state reflects the decision variable, and the coupling strength is determined by the group collaborative risk quantification. The local field strength is correlated with the consistency of the current gradient field direction; Optimize the population through a hybrid strategy of quantum tunneling mutation and dynamic reference point adjustment. The finally output Pareto optimal solution set is transformed into a continuous trajectory scheme empowered by the spatio-temporal gradient field, and its coefficients are directly mapped by the optimization variables. The heading instructions and reorganization marks inherit from the decision encoding of the optimal solution.

[0009] In a second aspect, the present application also provides a system for the emergence of a bionic swarm trajectory of unmanned aerial vehicles with a spatio-temporal gradient field, including: An acquisition module: used to acquire satellite remote sensing ice crack distribution data, LiDAR real-time terrain scanning data, ocean current velocity field data, surface temperature gradient data, and instantaneous wind speed vector data, and generate a four-dimensional spatio-temporal gradient field tensor matrix through a dynamic field coupling convolutional network; Building module: Used to construct a bionic group interaction rule set through non-linear differential game modeling based on the four-dimensional spatio-temporal gradient field tensor matrix, combined with the density-velocity response data of fish school escape behavior and the dynamic constraints of drones; Solving module: Used to utilize the bionic group interaction rule set and the four-dimensional spatio-temporal gradient field tensor matrix, adopt the spatio-temporal finite element discretization method and combine GPU accelerated parallel solution to determine the basis function of the individual drone trajectory; Processing module: Used to dynamically adjust the four-dimensional spatio-temporal gradient field tensor matrix according to the basis function of the individual drone trajectory and the real-time updated data of the environmental sensor, and obtain the updated spatio-temporal gradient field and generate a group recombination instruction through Lyapunov function optimization processing; Generating module: Used to synthesize the updated spatio-temporal gradient field, the group recombination instruction and the historical trajectory data, and apply the quantum annealing - NSGAIII hybrid optimization algorithm to generate a bionic group trajectory emergence scheme of drones empowered by the spatio-temporal gradient field.

[0010] Thirdly, the present application also provides a bionic group trajectory emergence device of drones for spatio-temporal gradient field, including: A memory, used to store computer programs; A processor, used to implement the steps of the bionic group trajectory emergence method of drones for spatio-temporal gradient field when executing the computer program.

[0011] Fourthly, the present application also provides a readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned bionic group trajectory emergence method of drones based on spatio-temporal gradient field are implemented.

[0012] The beneficial effects of the present invention are as follows: The present invention realizes millimeter-level perception of environmental threats by constructing a dynamic weighted model of the repulsion gradient at the edge of ice cracks through real-time fusion of satellite, LiDAR and buoy array data by a four-dimensional spatio-temporal gradient field tensor matrix (repulsion field, gravitational field, drift field); designs bionic constraints based on the density-velocity response curve of fish school escape behavior, combines Hamiltonian function optimization and spatio-temporal basis function expansion to generate distributed cooperation rules with low communication dependence; adopts quantum annealing to initialize the population and NSGA-III multi-objective search, and through risk classification GPU parallelism (double-precision calculation of 32×32 thread blocks in high-risk areas) and dynamic reference point adjustment, while ensuring the trajectory accuracy in high-risk areas (≤0.1m), it compresses the optimization time of a group of thousands of drones to within 5ms, overcoming the technical problem of autonomous cooperation of drone groups in polar dynamic environments.

[0013] Other features and advantages of the present invention will be described in the subsequent specification, and in part will be obvious from the specification, or can be understood by implementing the embodiments of the present invention. The objectives and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written specification, claims, and drawings. Brief Description of the Drawings

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0015] Figure 1 Schematic flow diagram of the method for the emergence of the trajectories of a bionic swarm of drones in a spatio-temporal gradient field described in the embodiments of the present invention; Figure 2 Schematic structural diagram of the system for the emergence of the trajectories of a bionic swarm of drones in a spatio-temporal gradient field described in the embodiments of the present invention; Figure 3 Schematic structural diagram of the device for the emergence of the trajectories of a bionic swarm of drones in a spatio-temporal gradient field described in the embodiments of the present invention.

[0016] In the figure: 701, acquisition module; 702, construction module; 703, solution module; 704, processing module; 705, generation module; 800, device for the emergence of the trajectories of a bionic swarm of drones in a spatio-temporal gradient field; 801, processor; 802, memory; 803, multimedia component; 804, I / O interface; 805, communication component. Detailed Description of the Embodiments

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Usually, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the present invention claimed, but only represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0018] It should be noted that similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of the present invention, terms such as "first" and "second" are only used for distinguishing descriptions and cannot be construed as indicating or implying relative importance.

[0019] Embodiment 1:

[0020] This embodiment provides a method for the emergence of the trajectories of a bionic group of unmanned aerial vehicles in a spatio-temporal gradient field.

[0021] See Figure 1 , which shows that this method includes step S100, step S200, step S300, step S400, and step S500.

[0022] S100. Obtain satellite remote sensing ice crack distribution data, LiDAR real-time terrain scanning data, ocean current velocity field data, surface temperature gradient data, and instantaneous wind speed vector data, and generate a four-dimensional spatio-temporal gradient field tensor matrix through a dynamic field coupling convolutional network.

[0023] It can be understood that in this step S100, it includes S101 and S102, where: S101. Obtain an ice surface image through a synthetic aperture radar carried by a polar orbiting satellite, extract a crack distribution binary map with a spatial resolution ≤ 1 m through an ice crack edge detection algorithm to obtain satellite remote sensing ice crack distribution data; collect original point cloud data through a solid-state LiDAR carried by an unmanned aerial vehicle, register multiple frames of point clouds through a point cloud registration algorithm, and generate a digital elevation model to obtain LiDAR real-time terrain scanning data; based on Doppler current meters built in a terrain sensor array with a spacing of 500 m, measure the horizontal velocity vector in real time, and generate an ocean current velocity field data through Kriging spatial interpolation; obtain an ice surface temperature distribution matrix at a sampling rate of 5 Hz through a high-resolution infrared thermal imager carried by an unmanned aerial vehicle, and calculate the spatial gradient field based on the Sobel operator to obtain surface temperature gradient data; obtain instantaneous wind speed vector data through Kriging interpolation calculation of the three-dimensional wind speed obtained by a complex environmental weather station network; It should be noted that for the satellite remote sensing ice crack distribution data: an ice surface image is obtained through a synthetic aperture radar (SAR) carried by a polar orbiting satellite, and a crack distribution binary map is extracted through an ice crack edge detection algorithm (based on the Canny operator and morphological closing operation), with a spatial resolution ≤ 1 m, updated daily; a solid-state LiDAR (wavelength 905 nm, scanning frequency 10 Hz) is carried by an unmanned aerial vehicle, and a digital elevation model (DEM) is generated through a point cloud registration algorithm, with an accuracy of ±0.1 m; a terrain sensor array (spacing 500 m) is deployed, and a Doppler current meter is built in to measure the horizontal velocity vector , sampling rate is 1 Hz; the drone is equipped with an infrared thermal imager (band 8 - 14 μm, thermal sensitivity ≤ 0.05 °C) to collect temperature distribution T ( x , y , t ), calculate the gradient field ; a complex environmental weather station network (node spacing 1 km) provides three-dimensional wind speed , and generates a continuous wind field through Kriging interpolation.

[0024] S102. Encode the satellite remote sensing ice crack distribution data, LiDAR real-time terrain scanning data, ocean current velocity field data, surface temperature gradient data, and instantaneous wind speed vector data into a five-channel spatio-temporal tensor, and use the five-channel spatio-temporal tensor as the input of the dynamic convolution network, thereby constructing a dynamic convolution neural network. The convolution kernel weights are jointly modulated by the ice crack curvature and temperature gradient, and the output is a coupled drift field vector. Combining the coordinates and priority weights provided by the task requirement system, generate a four-dimensional spatio-temporal gradient field tensor matrix, where the four-dimensional spatio-temporal gradient field tensor matrix includes a gravitational field, a repulsive field, and a drift field.

[0025] It should be noted that encoding the above data into a five-channel tensor , where is a binary map of ice crack distribution (0 / 1 represents safe / dangerous area); is the LiDAR elevation matrix; is the normalized physical field scalar or vector, and then design an ice crack curvature-temperature gradient joint modulation mechanism; among them, the convolution kernel is designed as: , where σ is the Sigmoid activation function, W is the trainable weight matrix, and the output is the dynamic coupling field Fdynamic ( x , y , t ). Among them, gravitational field: the priority weight of the target point generates a potential energy gradient, that is, according to the task material delivery target point and its priority, generate a time-varying gravitational field; repulsive field: the ice crack edge gradient (calculated based on the LiDAR point cloud curvature, and a repulsive peak is generated in the area where the curvature > threshold); drift field: the superposition of ocean current and wind speed vector generates a time-varying drift correction term (the dynamic Bayesian network predicts the field strength distribution in the next 10 s), that is, extract the multi-source field coupling characteristics through three-layer convolution (kernel size 3×3, stride 1), and output the dynamic drift field. Combine the repulsive force gradient, drift field, and gravitational gradient, that is, the formula is:

[0026] Among them, is the spatio-temporal gradient field tensor, is the potential field gradient, pointing to the steepest descent direction of the combined potential field (repulsion + attraction), is the drift field.

[0027] Among them, corrects the deficiencies of the traditional binary ice crack map through DEM elevation, avoiding misjudgment of the safe altitude of drones in slope terrain, while the weights of the DCNN convolution kernels are modulated by the temperature gradient ∇ T to respond in advance to the sudden change of ocean currents caused by ice melting on the ice surface. Moreover, the spatio-temporal gradient field G is not only the input of differential game but also optimizes its own parameters in reverse through trajectory data to form an adaptive control loop. Therefore, after obtaining the four-dimensional spatio-temporal gradient field tensor matrix, it provides a physical driving field for modeling bionic swarm rules.

[0028] S200. Based on the four-dimensional spatio-temporal gradient field tensor matrix, combined with the density-velocity response data of fish school escape behavior and the dynamic constraints of drones, construct a set of bionic swarm interaction rules through nonlinear differential game modeling.

[0029] It can be understood that in this step S200, it includes S201, S202, S203 and S204, where: S201. Based on the group density-velocity observation data of fish school escape behavior, use the Sigmoid function to fit the nonlinear mapping relationship between the density and velocity, and generate a bionic density-velocity response constraint curve. Its calculation formula is as follows:

[0030] In the formula, is the density-velocity response constraint curve, is the maximum allowable speed of the drone, is the drone swarm density, is the critical density threshold, k is the steepness coefficient of the Sigmoid curve, and e is the base of the natural logarithm, ; It should be noted that when a fish school encounters a predator, its escape speed and group density show a nonlinear relationship (the higher the density, the lower the individual speed limit to avoid collisions). Quantify this biological behavior law into mathematical constraint conditions and embed it into the drone differential game model to form a bionic differential constraint. Then this curve is the inequality constraint of the differential game, restricting the maximum speed of the drone in a dense group and preventing collisions or out-of-control caused by excessive speed during low-altitude flight in the polar region. Among them, the peak escape speed of the fish school ≤ 1.5 m / s, and fitting the fish school data gives k = 3.0, ensuring that the speed decays rapidly when ρ ∈ [1.8, 2.2] kg / m3.

[0031] S202. Construct a Hamiltonian function by combining the bionic density-velocity response constraint curve with a preset UAV dynamics model to generate a dynamically feasible solution space for the nonlinear differential game. The UAV dynamics model is obtained by integrating the measured polar parameters, gradient field coupling, and bionic rule constraints. The calculation formula for constructing the Hamiltonian function is as follows:

[0032] In the formula, is the UAV control input, is the Hamiltonian function, is the co-state vector, T is a matrix, is the UAV state equation, is the control input matrix; It should be noted that the construction of the UAV dynamics model includes: extracting the physical relationships describing position, velocity, and attitude from existing UAV dynamics theories, restricting the control input by measuring that the maximum thrust decreases from 40 N to 35 N in a 30°C environment, and measuring through polar experiments μ = 0.03 (only 1 / 10 of that on conventional ground). It is necessary to adjust the attitude control parameters, embed the obtained four-dimensional spatio-temporal gradient field tensor matrix into the dynamic equation to obtain environmental force coupling, and restrict the velocity through fish school rules. In the Hamiltonian function, the state equation f ( x ) includes the gradient field and the drift field , the control input satisfies , to obtain a dynamically feasible solution space for the nonlinear differential game and ensure that the trajectory satisfies the physical limits of the UAV.

[0033] S203. Based on the ice crack repulsive force gradient field component in the four-dimensional spatio-temporal gradient field tensor matrix, fuse the real-time relative positions of the UAV swarm, and generate a combined force vector field for group obstacle avoidance and collision avoidance through the inverse proportional repulsive potential superposition algorithm; It should be noted that in this step, it is to ensure that the UAVs can avoid ice cracks and avoid collisions with each other in the polar environment.

[0034] In this embodiment, this step S203 includes S2031, S2032, and S2033, where: S2031. Calculate the ice crack repulsive force gradient field covering the global mission area based on the ice crack repulsive force gradient field component in the four-dimensional spatio-temporal gradient field tensor matrix; S2032. Extract the local repulsive force gradient in the ice crack repulsive force gradient field, combine it with the real-time positions of the UAVs, and generate an anti-collision force between groups through inverse proportional potential superposition. The calculation formula is as follows:

[0035] Wherein, and represent the positions of UAVs i and j respectively, γ = 0.8 is the repulsive force weight coefficient, is the inter-group collision avoidance force received by the UAV i ; is the Euclidean distance between two UAVs; S2033. Superimpose the vectors of the local repulsive force gradient and the inter-group anti-collision force to obtain the total obstacle avoidance and anti-collision combined force of the UAV individual, and its calculation formula is as follows:

[0036] Wherein, is the local repulsive force gradient, is the inter-group anti-collision force, is the combined force vector field of group obstacle avoidance and anti-collision.

[0037] It should be noted that the repulsive force weight γ is adaptively adjusted according to the curvature of the ice crack to ensure the avoidance priority in high-risk areas (such as the edge of the ice crack). In this step, the edge curvature of the ice crack is used to dynamically adjust the repulsive force intensity: when > 0.7 (high-risk crack area), the repulsive force weight γ is automatically enhanced (such as γ → 1.2 γ ), forcing the UAV to detour more significantly; when ≤ 0.7 (low-risk area), the standard weight γ = 0.8 is adopted to balance obstacle avoidance and flight efficiency. The finally obtained Fsep is a vector force, and its direction is jointly determined by the ice crack repulsive force gradient and the repulsive force between UAVs, ensuring that the UAV can achieve autonomous obstacle avoidance and group collaboration in the polar complex environment, and enabling the UAV to intelligently identify and preferentially avoid high-risk crack areas during flight while maintaining a safe distance between groups.

[0038] S204. According to the data of the dynamic feasible solution space, the combined force vector field of group obstacle avoidance and anti-collision, and the drift field, through distributed Nash equilibrium solution and heading coordination optimization, obtain a set of bionic group interaction rules, including static constraints and dynamic coordination strategies.

[0039] It should be noted that the kinematically feasible solution space contains all candidate trajectory sets that satisfy polar dynamic constraints (such as low-temperature thrust limit, ice surface friction coefficient), and each trajectory is accompanied by multi-objective evaluation indicators such as energy consumption, time, and safety; the combined force vector field for group obstacle avoidance and collision avoidance, where the force direction at each point is dominated by the repulsion gradient of ice cracks and the intensity is dynamically adjusted according to the curvature (the intensity in high-risk areas is increased by 50%); and the data of the drift field contains the spatio-temporal variations of ocean currents and wind speeds, and its gradient can predict the trend of environmental disturbances in the next 10 seconds.

[0040] It can be understood that each UAV, as an independent game player, has a strategy space that is a subset of the kinematically feasible solutions (such as only retaining the candidate trajectories with the top 20% in terms of energy consumption).

[0041] Therefore, the main objective is: Then, weight allocation is carried out, and the asynchronous distributed ADMM algorithm is adopted. Each UAV iteratively optimizes its strategy based on neighbor information: that is, adjusts the candidate trajectory according to the current combined force; exchanges trajectory scores through a polar low-bandwidth channel (such as LoRa) to negotiate conflict segments; when the strategy change rate of all UAVs is < 1%, it is determined as a Nash equilibrium. The heading is aligned with that issued by the master UAV (based on the gravitational direction of the gradient field); if is detected, then the Brownian motion term compensation is activated, and the heading follows a stochastic differential equation; after the wind disturbance ends, the cooperation is quickly restored through the Lyapunov function. Therefore, through the prediction of the drift field gradient and local games, the task completion rate is still maintained at 80% when the communication packet loss rate is 30%, and the distributed solution reduces the computational load to 1 / 5 of the centralized one. The density-velocity relationship of fish school escape is encoded as a constraint condition; the drift field gradient is used to replace the biochemical signal transmission to achieve physically driven cooperation.

[0042] S300. Using the bionic group interaction rule set and the four-dimensional spatio-temporal gradient field tensor matrix, adopting the spatio-temporal finite element discretization method and combining with GPU-accelerated parallel solution, determine the trajectory basis function of the UAV individual.

[0043] It can be understood that in this step S300, it includes S301, S302, S303, and S304, where: S301. According to the four-dimensional spatio-temporal gradient field tensor matrix and the bionic group heading cooperation constraint, adopt the Fourier-Legendre combined basis function expansion method to construct a spatio-temporal threat perception basis function set. Its spatial wave number component is modulated by the ice crack distribution density, the time frequency component is correlated with the time-varying rate of the drift field, and the heading cooperation angle is embedded in the basis function phase through a cosine modulation term. Its calculation formula is as follows:

[0044] In the formula, is the kspatio-temporal coupled basis functions is the spatial wave term, representing the periodic change of the basis function on the horizontal plane is the spatial wave number modulated by the ice crack distribution is the time frequency parameter is the group heading coordination angle is the spatio-temporal gradient field tensor G is the modulus length of, reflecting the environmental threat intensity is the group heading coordination angle, and t is the time variable; S302. Discretize the spatio-temporal threat perception basis function set and the UAV dynamics model by the Galerkin weighted residual method to generate a trajectory coupling algebraic system where the coefficient matrix is composed of the inner product of the basis functions, and the right-hand side term contains the orthogonal projection of the group collision avoidance force;

[0045] In the formula, is the integral operation in the spatio-temporal domain is the UAV i is the partial derivative of the trajectory function with respect to time t is the negative gradient of the ice crack repulsive force field is the environmental drift field is the control input of the UAV is the k-th spatio-temporal basis function is the product of the spatial and time elements; S303. Based on the trajectory coupling algebraic system and the four-dimensional spatio-temporal gradient field tensor matrix, perform GPU parallel solution processing for ice gap risk grading to obtain a risk zoning strategy, where the risk zoning strategy includes a high-risk area with a 32×32 thread block allocated within 0.5 m of the ice crack edge, a medium-risk area with a 64×64 thread block allocated in the 1-2 m area outside the crack, and a safe area with a 128×128 thread block allocated in the flat ice surface area. The high-risk area uses double-precision floating-point operations and preconditioning acceleration, and the medium- and low-risk areas use single-precision hybrid iterative calculations; S304. Output the basis function coefficient vector through the risk zoning strategy, thereby generating the UAV continuous trajectory function.

[0046] It should be noted that the calculation resource priority is dynamically divided according to the ice crack curvature, and the calculation density in the high-risk area is increased by 3 times; when dynamically controlling the residual convergence, the real-time drift field change rate and the iterative residual sequence are input. When in the sudden disturbance mode, when the real-time drift field change rate is greater than , tighten the residual threshold to , sacrificing accuracy to ensure real-time performance; when the "chase-relax" iteration strategy is activated, the solution from the previous step is used for extrapolation to accelerate convergence. That is, when a sudden change in the drift field is detected, the residual threshold is tightened to and the chase-relax iteration is activated. It can be understood that in special cases, such as responding to polar sudden environmental events (such as snowstorms), a trade-off between computational accuracy and real-time performance is made through dynamic threshold switching. If the risk zones are skipped and direct homogenization is performed, the trajectory error in the high-risk area will increase by 2.3 times; if the preconditioner ignores the repulsive force gradient, the number of iterations at the ice crack edge will increase by 70%; if the convergence strategy is static, the solution failure rate under sudden disturbances will exceed 30%.

[0047] S400. According to the individual trajectory basis functions of the drones and the real-time updated data from the environmental sensors, through Lyapunov function optimization, the four-dimensional spatio-temporal gradient field tensor matrix is dynamically adjusted to obtain the updated spatio-temporal gradient field and generate a group reorganization instruction.

[0048] It can be understood that the calculation formula for the Lyapunov function optimization in this step S400 is as follows:

[0049] In the formula, L ( t ) is a scalar function representing the overall trajectory tracking error of the drone swarm, is the actual position of drone i, is the predicted position of drone i, is the actual velocity of drone i, is the predicted velocity of drone i, λ is the velocity error weight coefficient, and N is the size of the drone swarm.

[0050] It should be noted that the "dynamically adjusting the four-dimensional spatio-temporal gradient field tensor matrix to obtain the updated spatio-temporal gradient field and generate a group reorganization instruction" includes, after constructing the Lyapunov function, obtaining the overall stability index of the group, and using the quasi-Newton method to inversely optimize the gradient field parameters, that is:

[0051] In the formula, Calculated through the chain rule, correlating the trajectory error with the repulsive force gradient , if a sudden wind disturbance is detected , a vortex damping term is added to the drift field, and then the updated spatio-temporal gradient field is output G ′( x , y , z , t); and according to the updated spatiotemporal gradient field and group connectivity index, when the connectivity index is greater than 0.7, the task area is re-divided based on the Voronoi diagram to generate subgroup target points. The subgroup target points are generated by the repulsive gradient drive and combined with the drift field correction amount Generate a reorganization instruction set, which includes the forced alignment of the heading of drones in high-risk areas and the forced alignment of drones in low-risk areas. Drift field corrects heading.

[0052] S500, based on the updated space-time gradient field, group reorganization instructions and historical trajectory data, applies the quantum annealing-NSGAIII hybrid optimization algorithm to generate a bionic group trajectory emergence solution for drones enabled by the space-time gradient field.

[0053] It can be understood that step S500 includes S501, S502 and S503, wherein: S501. Based on the updated spatiotemporal gradient field, group reorganization instructions and historical trajectory data, a multi-objective problem including minimization of energy consumption, minimization of ice crack collision times and optimization of task synchronization is constructed, wherein the control input is constrained by the maximum thrust and bionic speed of the UAV, the ice crack collision judgment is triggered by the curvature field threshold, and the task synchronization is indexed by the arrival time extreme difference; S502. Generate an initial population using a quantum annealing mechanism, whose Hamiltonian is designed to couple historical trajectory conflict statistics with the alignment of the repulsive field gradient direction. Each individual encodes the control input sequence and heading strategy of the drone, where the quantum bit state reflects the decision variable, the coupling strength is determined by the group collaborative risk quantification, and the local field strength is correlated with the current gradient field direction consistency; It should be noted that the quantum annealing mechanism is used to generate the initial population, and its Hamiltonian is designed as:

[0054] In the formula, is the quantum bit state, is the coupling strength, The local field strength is positively correlated with the alignment of the repulsive gradient direction.

[0055] S503, through the hybrid strategy of quantum tunneling mutation and dynamic reference point adjustment to optimize the population, the final output Pareto optimal solution set is converted into a continuous trajectory scheme enabled by the space-time gradient field, whose coefficients are directly mapped by the optimization variables, and the heading instructions and reorganization marks are inherited from the decision coding of the optimal solution. The calculation formula is as follows:

[0056] In the formula, is the optimized UAV trajectory function, K is the number of basis functions, is the optimal basis function coefficient, are the horizontal position x and the mission time t, space-time basis functions.

[0057] The present invention realizes millimeter-level perception of environmental threats by constructing a dynamic weighted model of ice crack edge repulsive gradient through real-time fusion of satellite, LiDAR and buoy array data using a four-dimensional space-time gradient field tensor matrix; designs bionic constraints based on the density-velocity response curve of fish school escape behavior, combines Hamiltonian function optimization and space-time basis function expansion to generate distributed cooperation rules with low communication dependence; uses quantum annealing to initialize the population and NSGA-III multi-objective search, and compresses the optimization time of a thousand-level swarm to within 5 ms through risk-level GPU parallelism and dynamic reference point adjustment, overcoming the technical problem of autonomous cooperation of UAV swarms in polar dynamic environments.

[0058] Embodiment 2:

[0059] As Figure 2 shown, this embodiment provides a UAV bionic swarm trajectory emergence system for a space-time gradient field. Refer to Figure 2 The system includes: An acquisition module 701: used to acquire satellite remote sensing ice crack distribution data, LiDAR real-time terrain scanning data, ocean current velocity field data, surface temperature gradient data and instantaneous wind speed vector data, and generate a four-dimensional space-time gradient field tensor matrix through a dynamic field coupling convolutional network; A construction module 702: used to construct a bionic swarm interaction rule set through non-linear differential game modeling based on the four-dimensional space-time gradient field tensor matrix, combined with the density-velocity response data of fish school escape behavior and the dynamic constraints of UAVs; A solution module 703: used to determine the UAV individual trajectory basis function by using the bionic swarm interaction rule set and the four-dimensional space-time gradient field tensor matrix, adopting a space-time finite element discretization method and combining GPU accelerated parallel solution; A processing module 704: used to dynamically adjust the four-dimensional space-time gradient field tensor matrix through Lyapunov function optimization processing according to the UAV individual trajectory basis function and the real-time updated data of the environmental sensor, obtain the updated space-time gradient field and generate a swarm reorganization instruction; A generation module 705: used to generate a UAV bionic swarm trajectory emergence scheme empowered by the space-time gradient field by applying a quantum annealing-NSGAIII hybrid optimization algorithm, integrating the updated space-time gradient field, the swarm reorganization instruction and the historical trajectory data.

[0060] Specifically, the acquisition module 701 includes: Acquisition Unit: It is used to obtain ice surface images through the synthetic aperture radar carried by polar-orbiting satellites, extract the binary map of crack distribution with a spatial resolution ≤ 1m through the ice crack edge detection algorithm, and obtain satellite remote sensing ice crack distribution data; collect original point cloud data through the solid-state lidar carried by the unmanned aerial vehicle (UAV), register multiple frames of point clouds through the point cloud registration algorithm, generate a digital elevation model, and obtain LiDAR real-time terrain scanning data; based on the Doppler current meters built in the terrain sensor array with a spacing of 500m, measure the horizontal velocity vector in real time, and generate ocean current velocity field data through Kriging spatial interpolation; through the high-resolution infrared thermal imager carried by the UAV, obtain the ice surface temperature distribution matrix at a sampling rate of 5Hz, and calculate the spatial gradient field based on the Sobel operator to obtain surface temperature gradient data; through Kriging interpolation calculation of the three-dimensional wind speed obtained by the complex environmental weather station network, obtain instantaneous wind speed vector data; First Construction Unit: It is used to encode the satellite remote sensing ice crack distribution data, LiDAR real-time terrain scanning data, ocean current velocity field data, surface temperature gradient data and instantaneous wind speed vector data into a five-channel spatio-temporal tensor, and use the five-channel spatio-temporal tensor as the input of the dynamic convolutional network, so as to construct a dynamic convolutional neural network. The convolution kernel weights are jointly modulated by the ice crack curvature and temperature gradient, and the output is a coupled drift field vector. Combining the coordinates and priority weights provided by the task requirement system, a four-dimensional spatio-temporal gradient field tensor matrix is generated, where the four-dimensional spatio-temporal gradient field tensor matrix includes a gravitational field, a repulsive field and a drift field.

[0061] Specifically, the construction module 702, which includes: First Generation Unit: It is used to fit the non-linear mapping relationship between the density and velocity based on the population density-velocity observation data of fish school escape behavior, and generate a bionic density-velocity response constraint curve. Its calculation formula is as follows:

[0062] In the formula, is the density-velocity response constraint curve, is the maximum allowable speed of the UAV, is the UAV population density, is the critical density threshold, k is the steepness coefficient of the Sigmoid curve, and e is the base of the natural logarithm, ; Second Construction Unit: It is used to construct the Hamiltonian function by combining the bionic density-velocity response constraint curve with the preset UAV dynamics model, and generate the dynamic feasible solution space of the non-linear differential game. The UAV dynamics model includes those obtained by fusing the measured polar parameters, gradient field coupling and bionic rule constraints. The calculation formula for constructing its Hamiltonian function is as follows:

[0063] In the formula, is the UAV control input, is the Hamiltonian function, is the co-state vector, and T is a matrix, is the UAV state equation, is the control input matrix; The second generation unit: is used to fuse the real-time relative positions of the UAV swarm based on the ice crack repulsive force gradient field component in the four-dimensional spatio-temporal gradient field tensor matrix, and generate a combined force vector field for group obstacle avoidance and collision avoidance through the inverse proportional repulsive potential superposition algorithm; The optimization unit: is used to obtain a set of bionic group interaction rules, including static constraints and dynamic cooperation strategies, through distributed Nash equilibrium solution and heading coordination optimization according to the data of the dynamic feasible solution space, the combined force vector field for group obstacle avoidance and collision avoidance, and the drift field.

[0064] Specifically, the second generation unit includes: The first calculation unit: is used to calculate the ice crack repulsive force gradient field covering the global task area based on the ice crack repulsive force gradient field component in the four-dimensional spatio-temporal gradient field tensor matrix; The second calculation unit: is used to extract the local repulsive force gradient in the ice crack repulsive force gradient field, and combine it with the real-time position of the UAV to generate an anti-collision force between groups through inverse proportional potential superposition. The calculation formula is as follows:

[0065] In the formula, and respectively represent the positions of UAVs i and j, γ = 0.8 is the repulsive force weight coefficient, is the UAV i the anti-collision force between groups received, the Euclidean distance between two UAVs; The third calculation unit: is used to superimpose the vectors of the local repulsive force gradient and the anti-collision force between groups to obtain the total obstacle avoidance and collision avoidance combined force of the UAV individual. The calculation formula is as follows:

[0066] In the formula, is the local repulsive force gradient, is the anti-collision force between groups, is the combined force vector field for group obstacle avoidance and collision avoidance.

[0067] Specifically, the solving module 703 includes: Constructor function set unit: Used to construct a spatio-temporal threat perception basis function set according to the four-dimensional spatio-temporal gradient field tensor matrix and the bionic group heading collaborative constraint, adopting the Fourier-Legendre combined basis function expansion method. Its spatial wave number component is modulated by the ice crack distribution density, the time frequency component is correlated with the time-varying rate of the drift field, and the heading collaborative angle is embedded in the basis function phase through a cosine modulation term. Its calculation formula is as follows:

[0068] In the formula, is the k -th spatio-temporal coupling basis function, is the spatial wave term, representing the periodic change of the basis function in the horizontal plane, is the spatial wave number modulated by the ice crack distribution, is the time frequency parameter, is the group heading collaborative angle, is the spatio-temporal gradient field tensor G 's modulus length, reflecting the environmental threat intensity, is the group heading collaborative angle, and t is the time variable; Discrete generation unit: Used to discretize the spatio-temporal threat perception basis function set and the UAV dynamics model through the Galerkin weighted residual method to generate a trajectory coupling algebraic system , where the coefficient matrix is composed of the inner product of the basis functions, and the right-hand side term contains the orthogonal projection of the group collision avoidance force;

[0069] In the formula, is the integral operation in the spatio-temporal domain, is the partial derivative of the trajectory function of the UAV i with respect to time t, is the negative gradient of the ice crack repulsive force field, is the environmental drift field, is the control input of the UAV, is the k-th spatio-temporal basis function, is the product of the spatial and time micro-elements; Partitioning strategy unit: Used to implement GPU parallel solution processing for ice gap risk grading based on the trajectory coupling algebraic system and the four-dimensional spatio-temporal gradient field tensor matrix to obtain a risk partitioning strategy. The risk partitioning strategy includes a high-risk area within 0.5 m of the ice crack edge, where 32×32 thread blocks are allocated, a medium-risk area in the 1-2 m area outside the crack, where 64×64 thread blocks are allocated, and a safe area on the flat ice surface, where 128×128 thread blocks are allocated. The high-risk area uses double-precision floating-point operations and preconditioning acceleration, and the medium- and low-risk areas use single-precision hybrid iterative calculations; Output generation unit: It is used to output the basis function coefficient vector through the risk zoning strategy, so as to generate the continuous trajectory function of the UAV.

[0070] Specifically, the generation module 705 includes: The third construction unit: It is used to construct a multi-objective problem including minimizing energy consumption, minimizing the number of ice crack collisions, and optimizing task synchronization based on the updated spatio-temporal gradient field, population recombination instruction, and historical trajectory data. Among them, the control input is restricted by the maximum thrust and bionic speed of the UAV, the ice crack collision determination is triggered by the curvature field threshold, and the task synchronization is measured by the range of arrival times; The third generation unit: It is used to generate the initial population using the quantum annealing mechanism. The Hamiltonian design couples the historical trajectory conflict statistics and the alignment degree of the repulsive force field gradient direction. Each individual encodes the control input sequence and heading strategy of the UAV. Among them, the qubit state reflects the decision variable, the coupling strength is determined by the population collaborative risk quantification, and the local field strength is correlated with the consistency of the current gradient field direction; The transformation unit: It is used to optimize the population through a hybrid strategy of quantum tunneling mutation and dynamic reference point adjustment. The finally output Pareto optimal solution set is transformed into a continuous trajectory scheme empowered by the spatio-temporal gradient field. Its coefficients are directly mapped by the optimization variables, and the heading instruction and recombination mark inherit the decision encoding of the optimal solution. The calculation formula is as follows:

[0071] In the formula, is the optimized UAV trajectory function, K is the number of basis functions, is the optimal basis function coefficient, is the horizontal position x and the mission time t, Spatio-temporal basis function.

[0072] It should be noted that regarding the system in the above embodiments, the specific ways in which each module performs operations have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0073] Embodiment 3:

[0074] Corresponding to the above method embodiment, in this embodiment, a bionic swarm trajectory emergence device of a UAV with a spatio-temporal gradient field is also provided. A bionic swarm trajectory emergence device of a UAV with a spatio-temporal gradient field described below can be mutually corresponding and referred to with a bionic swarm trajectory emergence method of a UAV with a spatio-temporal gradient field described above.

[0075] Figure 3 It is a block diagram of a bionic swarm trajectory emergence device 800 of a UAV with a spatio-temporal gradient field shown according to an exemplary embodiment. As Figure 3As shown, the UAV bionic swarm trajectory emergence device 800 of the spatio-temporal gradient field includes: a processor 801 and a memory 802. The UAV bionic swarm trajectory emergence device 800 of the spatio-temporal gradient field also includes one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0076] Among them, the processor 801 is used to control the overall operation of the bionic swarm trajectory emergence device 800 of the spatio-temporal gradient field to complete all or part of the steps in the above-mentioned method for the bionic swarm trajectory emergence of the spatio-temporal gradient field. The memory 802 is used to store various types of data to support the operation of the bionic swarm trajectory emergence device 800 of the spatio-temporal gradient field. These data may include, for example, instructions for any application or method operating on the bionic swarm trajectory emergence device 800 of the spatio-temporal gradient field, as well as application-related data, such as contact data, sent and received messages, pictures, audio, video, and so on. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic memory, flash memory, a magnetic disk or an optical disc. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 802 or sent through the communication component 805. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, and the above-mentioned other interface modules may be a keyboard, a mouse, or buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the bionic swarm trajectory emergence device 800 of the spatio-temporal gradient field and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination of one or more of them. Therefore, the corresponding communication component 805 may include: a Wi-Fi module, a Bluetooth module, or an NFC module.

[0077] In an exemplary embodiment, the UAV bionic swarm trajectory emergence device 800 of the spatio-temporal gradient field can be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components, and is used to execute the above-mentioned UAV bionic swarm trajectory emergence method of the spatio-temporal gradient field.

[0078] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided. When the program instructions are executed by a processor, the steps of the above-mentioned UAV bionic swarm trajectory emergence method of the spatio-temporal gradient field are implemented. For example, the computer-readable storage medium can be the above-mentioned memory 802 including program instructions, and the above program instructions can be executed by the processor 801 of the UAV bionic swarm trajectory emergence device 800 of the spatio-temporal gradient field to complete the above-mentioned UAV bionic swarm trajectory emergence method of the spatio-temporal gradient field.

[0079] Embodiment 4:

[0080] Corresponding to the above method embodiment, a readable storage medium is also provided in this embodiment. A readable storage medium described below can be correspondingly referred to with a UAV bionic swarm trajectory emergence method of a spatio-temporal gradient field described above.

[0081] A computer program is stored on the readable storage medium. When the computer program is executed by a processor, the steps of the UAV bionic swarm trajectory emergence method of the above method embodiment of the spatio-temporal gradient field are implemented.

[0082] The readable storage medium can specifically be various readable storage media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disc that can store program codes.

[0083] In summary, by fusing satellite, LiDAR, and buoy array data in real-time through the four-dimensional space-time gradient field tensor matrix (repulsion field, gravitational field, drift field), a dynamic weighted model of the repulsion gradient at the ice crack edge is constructed to achieve millisecond-level updates of environmental threats; the density-velocity response curve is incorporated into the Hamiltonian function to generate distributed rules that balance biological rationality and kinetic feasibility, reducing the communication load to 0.5 Mbps. Using the combined Fourier-Legendre basis function expansion and parallel computing by risk grading, the solution time for the trajectories of a thousand-level swarm is compressed to within 5 ms, and the accuracy in the high-risk area reaches the 0.1 m level; the density of reference points in the high-risk area is increased by 3 times, and combined with quantum tunneling mutation, the local optimum is broken through, and the Pareto solution set covers all dimensions of energy consumption, safety, and synchronization, with the task success rate increased to 98.7%. This provides high-precision, low-latency, and strong-robustness autonomous navigation capabilities for polar unmanned aerial vehicle swarms, significantly superior to the existing technical level.

[0084] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, various modifications and changes can be made to the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

[0085] As described above, this is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present invention, and all should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for the emergence of the trajectories of a bionic swarm of drones in a spatio-temporal gradient field, characterized in that, Including: Obtain satellite remote sensing ice crack distribution data, LiDAR real-time terrain scanning data, ocean current velocity field data, surface temperature gradient data, and instantaneous wind speed vector data, and generate a four-dimensional spatio-temporal gradient field tensor matrix through a dynamic field coupling convolutional network; Based on the four-dimensional spatio-temporal gradient field tensor matrix, combined with the density-velocity response data of fish swarm escape behavior and the dynamic constraints of unmanned aerial vehicles (UAVs), construct a bionic group interaction rule set through non-linear differential game modeling; Utilize the bionic group interaction rule set and the four-dimensional spatio-temporal gradient field tensor matrix, adopt the spatio-temporal finite element discretization method and combine GPU-accelerated parallel solution to determine the UAV individual trajectory basis function; According to the UAV individual trajectory basis function and the real-time updated data of the environmental sensor, through Lyapunov function optimization processing, dynamically adjust the four-dimensional spatio-temporal gradient field tensor matrix, obtain the updated spatio-temporal gradient field, and generate a group recombination instruction; Integrate the updated spatio-temporal gradient field, group recombination instruction, and historical trajectory data, apply the quantum annealing-NSGAIII hybrid optimization algorithm, and generate a UAV bionic group trajectory emergence scheme empowered by the spatio-temporal gradient field.

2. The method for the emergence of the trajectory of a bionic group of unmanned aerial vehicles in a spatio-temporal gradient field according to claim 1, characterized in that, The obtaining of satellite remote sensing ice crack distribution data, LiDAR real-time terrain scanning data, ocean current velocity field data, surface temperature gradient data, and instantaneous wind speed vector data, and generating a four-dimensional spatio-temporal gradient field tensor matrix through a dynamic field coupling convolutional network, including: Obtain ice surface images through the synthetic aperture radar carried by a polar orbiting satellite, extract a binary map of crack distribution with a spatial resolution ≤ 1m through an ice crack edge detection algorithm to obtain satellite remote sensing ice crack distribution data; collect raw point cloud data through a solid-state lidar carried by a UAV, register multiple frames of point clouds through a point cloud registration algorithm, and generate a digital elevation model to obtain LiDAR real-time terrain scanning data; based on the Doppler current meters built in a terrain sensor array with a spacing of 500m, measure the horizontal velocity vector in real time, and generate ocean current velocity field data through Kriging spatial interpolation; obtain the ice surface temperature distribution matrix at a sampling rate of 5Hz through a high-resolution infrared thermal imager carried by a UAV, and calculate the spatial gradient field based on the Sobel operator to obtain surface temperature gradient data; obtain instantaneous wind speed vector data through Kriging interpolation calculation of the three-dimensional wind speed obtained by a complex environmental weather station network; Encode the satellite remote sensing ice crack distribution data, LiDAR real-time terrain scanning data, ocean current velocity field data, surface temperature gradient data, and instantaneous wind speed vector data into a five-channel spatio-temporal tensor, and use the five-channel spatio-temporal tensor as the input of the dynamic convolutional network, thereby constructing a dynamic convolutional neural network, the convolution kernel weights of which are jointly modulated by the ice crack curvature and temperature gradient, the output is a coupled drift field vector, and combined with the coordinates and priority weights provided by the task requirement system, generate a four-dimensional spatio-temporal gradient field tensor matrix, where the four-dimensional spatio-temporal gradient field tensor matrix includes a gravitational field, a repulsive field, and a drift field.

3. The method for the emergence of the bionic swarm trajectory of drones in the spatio-temporal gradient field according to claim 1, characterized in that The constructing of a bionic group interaction rule set through non-linear differential game modeling based on the four-dimensional spatio-temporal gradient field tensor matrix, combined with the density-velocity response data of fish swarm escape behavior and the dynamic constraints of UAVs, including: Based on the group density-velocity observation data of fish swarm escape behavior, the Sigmoid function is used to fit the nonlinear mapping relationship between the density and the velocity, and a bionic density-velocity response constraint curve is generated. The calculation formula is as follows: In the formula, is the density-velocity response constraint curve, is the maximum allowable speed of the UAV, is the density of the UAV swarm, is the critical density threshold, k is the steepness coefficient of the Sigmoid curve, and e is the base of the natural logarithm, ; The bionic density-velocity response constraint curve and the preset UAV dynamics model are used to construct the Hamiltonian function, and a dynamic feasible solution space of the nonlinear differential game is generated. The UAV dynamics model includes those obtained by integrating the measured polar parameters, gradient field coupling, and bionic rule constraints. The calculation formula for constructing the Hamiltonian function is as follows: In the formula, is the UAV control input, is the Hamiltonian function, is the co-state vector, T is a matrix, is the UAV state equation, is the control input matrix; Based on the ice crack repulsive force gradient field component in the four-dimensional spatio-temporal gradient field tensor matrix, integrating the real-time relative positions of the UAV swarm, a combined force vector field for group obstacle avoidance and collision avoidance is generated through the inverse proportional repulsive potential superposition algorithm; According to the dynamic feasible solution space, the combined force vector field for group obstacle avoidance and collision avoidance, and the data of the drift field, through distributed Nash equilibrium solution and heading collaborative optimization, a bionic group interaction rule set is obtained, including static constraints and dynamic collaborative strategies.

4. The method for the emergence of the trajectories of a bionic group of drones in a spatio-temporal gradient field according to claim 3, characterized in that, The method of generating the combined force vector field for group obstacle avoidance and collision avoidance based on the ice crack repulsive force gradient field component in the four-dimensional spatio-temporal gradient field tensor matrix, integrating the real-time relative positions of the UAV swarm, through the inverse proportional repulsive potential superposition algorithm, includes: Based on the ice crack repulsive force gradient field component in the four-dimensional spatio-temporal gradient field tensor matrix, calculate the ice crack repulsive force gradient field covering the global mission area; Extract the local repulsive force gradient in the ice crack repulsive force gradient field, and combine it with the real-time position of the UAV to generate the anti-collision force between groups through the inverse proportional potential superposition. The calculation formula is as follows: In the formula, and respectively represent the positions of UAVs i and j, γ = 0.8 is the repulsive force weight coefficient, is the inter-group collision avoidance force received by UAV i , the Euclidean distance between two UAVs; Superimpose the vectors of the local repulsive force gradient and the anti-collision force between groups to obtain the total anti-obstacle and anti-collision combined force of the UAV individual. The calculation formula is as follows: In the formula, is the local repulsive force gradient, is the anti-collision force between groups, is the combined force vector field of group obstacle avoidance and collision avoidance.

5. The method for the emergence of the trajectory of a bionic swarm of drones in a spatio-temporal gradient field according to claim 3, wherein Using the bionic group interaction rule set and the four-dimensional spatio-temporal gradient field tensor matrix, adopting the spatio-temporal finite element discretization method and combining GPU accelerated parallel solution to determine the trajectory basis function of the UAV individual, including: According to the four-dimensional spatio-temporal gradient field tensor matrix and the bionic group heading collaborative constraint, adopt the Fourier-Legendre combined basis function expansion method to construct a spatio-temporal threat perception basis function set. Its spatial wave number component is modulated by the ice crack distribution density, the time frequency component is correlated with the time-varying rate of the drift field, and the heading collaborative angle is embedded in the basis function phase through the cosine modulation term. The calculation formula is as follows: In the formula, is the k th spatio-temporal coupling basis function, is the spatial fluctuation term, representing the periodic change of the basis function on the horizontal plane, is the spatial wave number modulated by the ice crack distribution, is the time frequency parameter, is the group heading coordination angle, is the spatio-temporal gradient field tensor G 's modulus length, reflecting the environmental threat intensity, is the group heading coordination angle, and t is the time variable; Discretize the spatio-temporal threat perception basis function set and the UAV dynamics model by the Galerkin weighted residual method to generate a trajectory-coupled algebraic system , where the coefficient matrix is composed of the inner product of the basis functions, and the right-hand side term contains the orthogonal projection of the group collision avoidance force. The calculation formula for the Galerkin weighted residual method discretization is as follows: In the formula, is the integral operation in the spatio-temporal domain, is the partial derivative of the trajectory function of the unmanned aerial vehicle i with respect to time t, is the negative gradient of the ice crack repulsive force field, is the environmental drift field, is the control input of the unmanned aerial vehicle, is the k-th spatio-temporal basis function, is the product of the spatial and temporal micro-elements; Based on the trajectory coupling algebraic system and the four-dimensional spatio-temporal gradient field tensor matrix, implement the GPU parallel solution processing for ice gap risk grading to obtain the risk zoning strategy. The risk zoning strategy includes a high-risk area with a 32×32 thread block allocated within 0.5 m of the ice crack edge, a medium-risk area with a 64×64 thread block allocated in the 1 - 2 m area outside the crack, and a safe area with a 128×128 thread block allocated in the flat ice surface area. The high-risk area uses double-precision floating-point operations and preconditioning acceleration, and the medium- and low-risk areas use single-precision hybrid iterative calculations; Output the basis function coefficient vector through the risk zoning strategy, thereby generating the continuous trajectory function of the UAV.

6. The method for the emergence of the trajectory of a bionic group of drones in a spatio-temporal gradient field according to claim 1, characterized in that The data is updated in real time according to the individual trajectory basis function of the unmanned aerial vehicle (UAV) and the environmental sensors, and is optimized by the Lyapunov function. The calculation formula for the optimization by the Lyapunov function is as follows: wherein, L ( t ) is a scalar function characterizing the overall trajectory tracking error of the UAV swarm, is the actual position of UAV i, is the predicted position of UAV i, is the actual velocity of UAV i, is the predicted velocity of UAV i, λ is the velocity error weight coefficient, and N is the scale of the UAV swarm.

7. The method for the emergence of the trajectories of a bionic swarm of drones in a spatio-temporal gradient field according to claim 1, wherein The comprehensively updated spatio-temporal gradient field, swarm reorganization instructions, and historical trajectory data are applied to the quantum annealing-NSGAIII hybrid optimization algorithm to generate a UAV bionic swarm trajectory emergence scheme empowered by the spatio-temporal gradient field, including: Based on the updated spatio-temporal gradient field, swarm reorganization instructions, and historical trajectory data, a multi-objective problem including minimizing energy consumption, minimizing the number of ice crack collisions, and optimizing task synchronization is constructed. Among them, the control input is restricted by the maximum thrust of the UAV and the bionic speed constraint. The ice crack collision determination is triggered by the curvature field threshold, and the task synchronization is measured by the range of arrival times; The quantum annealing mechanism is used to generate the initial population. The Hamiltonian design couples the historical trajectory conflict statistics and the alignment degree of the repulsive force field gradient direction. Each individual encodes the control input sequence and heading strategy of the UAV. Among them, the qubit state reflects the decision variable, the coupling strength is determined by the quantification of the swarm cooperation risk, and the local field strength is correlated with the consistency of the current gradient field direction; The population is optimized by a hybrid strategy of quantum tunneling mutation and dynamic reference point adjustment. The finally output Pareto optimal solution set is transformed into a continuous trajectory scheme empowered by the spatio-temporal gradient field. Its coefficients are directly mapped by the optimization variables. The heading instruction and reorganization mark inherit the decision encoding of the optimal solution. The calculation formula is as follows: In the formula, is the optimized UAV trajectory function, K is the number of basis functions, is the optimal basis function coefficient, are the horizontal position x and the mission time t, Space-time basis function.

8. A UAV bionic swarm trajectory emergence system for a spatio-temporal gradient field, based on the UAV bionic swarm trajectory emergence method for a spatio-temporal gradient field according to claim 1, characterized in that, Including: An acquisition module: used to acquire satellite remote sensing ice crack distribution data, LiDAR real-time terrain scanning data, ocean current velocity field data, surface temperature gradient data, and instantaneous wind speed vector data, and generate a four-dimensional spatio-temporal gradient field tensor matrix through a dynamic field coupling convolutional network; A construction module: used to construct a bionic swarm interaction rule set through non-linear differential game modeling based on the four-dimensional spatio-temporal gradient field tensor matrix, combined with the density-velocity response data of the fish school escape behavior and the UAV dynamics constraints; A solution module: used to determine the individual trajectory basis function of the UAV by using the bionic swarm interaction rule set and the four-dimensional spatio-temporal gradient field tensor matrix, adopting the spatio-temporal finite element discretization method and combining with GPU accelerated parallel solution; A processing module: used to dynamically adjust the four-dimensional spatio-temporal gradient field tensor matrix according to the individual trajectory basis function of the UAV and the real-time updated data of the environmental sensors, optimize through the Lyapunov function, obtain the updated spatio-temporal gradient field, and generate swarm reorganization instructions; A generation module: used to comprehensively apply the updated spatio-temporal gradient field, swarm reorganization instructions, and historical trajectory data to the quantum annealing-NSGAIII hybrid optimization algorithm to generate a UAV bionic swarm trajectory emergence scheme empowered by the spatio-temporal gradient field.

9. The UAV bionic swarm trajectory emergence system of the spatio-temporal gradient field according to claim 8, characterized in that, The acquisition module, including: Acquisition Unit: It is used to obtain ice surface images through the synthetic aperture radar carried by polar-orbiting satellites, extract a binary map of crack distribution with a spatial resolution ≤ 1m through an ice crack edge detection algorithm to obtain satellite remote sensing ice crack distribution data; collect original point cloud data through the solid-state lidar carried by the unmanned aerial vehicle (UAV), register multiple frames of point clouds through a point cloud registration algorithm to generate a digital elevation model, and obtain LiDAR real-time terrain scanning data; based on the Doppler current meters built in the terrain sensor array with a spacing of 500m, measure the horizontal velocity vector in real time, and generate ocean current velocity field data through Kriging spatial interpolation; obtain the ice surface temperature distribution matrix at a sampling rate of 5Hz through the high-resolution infrared thermal imager carried by the UAV, and calculate the spatial gradient field based on the Sobel operator to obtain surface temperature gradient data; obtain instantaneous wind speed vector data through Kriging interpolation calculation of the three-dimensional wind speed obtained from the complex environmental weather station network. First Construction Unit: It is used to encode the satellite remote sensing ice crack distribution data, LiDAR real-time terrain scanning data, ocean current velocity field data, surface temperature gradient data, and instantaneous wind speed vector data into a five-channel spatio-temporal tensor, and use the five-channel spatio-temporal tensor as the input of the dynamic convolutional network, thereby constructing a dynamic convolutional neural network. The convolution kernel weights are jointly modulated by the ice crack curvature and temperature gradient, and the output is a coupled drift field vector. Combining the coordinates and priority weights provided by the task requirement system, a four-dimensional spatio-temporal gradient field tensor matrix is generated, where the four-dimensional spatio-temporal gradient field tensor matrix includes a gravitational field, a repulsive field, and a drift field.

10. The bionic swarm trajectory emergence system of the spatio-temporal gradient field according to claim 8, characterized in that, The construction module, which includes: First Generation Unit: It is used to fit the non-linear mapping relationship between the density and velocity based on the population density-velocity observation data of the fish school escape behavior using the Sigmoid function, and generate a bionic density-velocity response constraint curve. Its calculation formula is as follows: In the formula, is the density-velocity response constraint curve, is the maximum allowable speed of the UAV, is the density of the UAV swarm, is the critical density threshold, k is the steepness coefficient of the Sigmoid curve, and e is the base of the natural logarithm, ; Second Construction Unit: It is used to construct a Hamiltonian function by combining the bionic density-velocity response constraint curve with a preset UAV dynamics model to generate a dynamically feasible solution space for non-linear differential games. The UAV dynamics model is obtained by integrating measured polar parameters, gradient field coupling, and bionic rule constraints. The calculation formula for constructing its Hamiltonian function is as follows: In the formula, is the UAV control input, is the Hamiltonian function, is the co-state vector, T is the matrix, is the UAV state equation, is the control input matrix; Second Generation Unit: It is used to generate a combined force vector field for group obstacle avoidance and collision avoidance based on the ice crack repulsive gradient field component in the four-dimensional spatio-temporal gradient field tensor matrix and by fusing the real-time relative positions of the UAV swarm through the inverse repulsive potential superposition algorithm. Optimization Unit: It is used to obtain a bionic group interaction rule set containing static constraints and dynamic cooperation strategies through distributed Nash equilibrium solution and heading coordination optimization based on the dynamically feasible solution space, the combined force vector field for group obstacle avoidance and collision avoidance, and the data of the drift field.

Images