Unmanned aerial vehicle cluster collaborative operation path planning method based on distributed computing power

Abstract

The invention discloses an unmanned aerial vehicle cluster collaborative operation path planning method based on distributed computing power. The method comprises the following steps: constructing a space-time safety corridor composed of convex polyhedrons for each unmanned aerial vehicle; an initial connected track is generated in the corridor, invariant re-parameterization based on curvature and torsion is carried out, and optimization solution is carried out in an invariant space; the result is expressed as a curvature continuous spline curve, a target function containing bending energy and path cost is constructed with control points and node time as variables, multiple constraints are applied, and complementary curvature constraints are established; minimizing the target function under constraint to obtain a smooth track; when an obstacle or a near conflict is detected, local re-optimization is carried out; and the edge nodes are responsible for constraint synchronization, independent optimization of each unmanned aerial vehicle and rolling updating of the flight path. According to the method, the stability of path planning and the energy utilization efficiency are improved by combining distributed computing power collaboration, geometric invariant re-parameterization and complementary curvature constraint.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of path planning, in particular to a method for path planning of cooperative operation of a UAV cluster based on distributed computing power. BACKGROUND

[0002] Cooperative operation of a UAV cluster is an important research direction in the field of intelligent equipment and air-ground cooperative control in recent years. With the wide application of UAVs in tasks such as logistics transportation, disaster inspection, emergency rescue and regional mapping, cluster flight scheduling has become a key technical means to improve task efficiency and operation coverage.

[0003] In the prior art, the path planning of a UAV is mostly in a centralized control or single-machine autonomous planning mode. Centralized path planning usually calculates the cluster flight path uniformly by a ground control center, although a global optimal solution can be obtained, but in the case of limited communication link or dynamic change of task environment, delay and failure problems are easy to occur; single-machine autonomous planning relies on local sensors and computing power, which can achieve local obstacle avoidance to a certain extent, but lacks global coordination, which is easy to cause path conflict or uneven resource allocation.

[0004] For multi-target task planning of a UAV group, researchers have proposed various methods based on graph search, dynamic programming, reinforcement learning and convex optimization. For example, path search based on A* or D* algorithm can quickly solve the shortest path in discrete space, but it is difficult to consider dynamic constraints and energy cost; continuous trajectory optimization methods based on Bezier curve, B-spline or polynomial curve can generate smooth flight paths, but when the constraint conditions are complex or there is interaction between multiple machines, the computational amount increases significantly and real-time performance is insufficient.

[0005] To solve the problem of calculation delay, some schemes introduce a distributed computing power architecture, which allocates trajectory optimization tasks to edge nodes or on-board processing units. However, most of the existing distributed planning still represents the trajectory in the form of traditional time or arc length parameterization, lacks modeling of trajectory geometric invariants such as curvature and torsion, making the optimization space highly dependent and the convergence speed limited. At the same time, the existing multi-UAV cooperative constraints often only rely on position or distance threshold judgment, without establishing a group coordination mechanism at the curvature level, which is easy to cause local aggregation or trajectory intersection in complex scenarios.

[0006] Therefore, how to provide a method for path planning of cooperative operation of a UAV cluster based on distributed computing power is a problem that those skilled in the art need to solve. SUMMARY

[0007] An object of the present application is to propose a UAV cluster cooperative operation path planning method based on distributed computing power. The present application establishes a double-layer solving mechanism under a distributed computing power architecture, divides trajectory planning into edge node and airborne end cooperative execution: the edge node is responsible for the synchronous solving and data exchange of multi-UAV constraints, the airborne end independently completes the spline optimization and constraint correction, forms a parallel computing structure to improve real-time performance and scalability; a trajectory reparameterization method based on geometric invariants is used to convert the flight path from spatial coordinates to invariant form of curvature and torsion, and optimization is performed in the invariant space to reduce parameter correlation and enhance curvature continuity and dynamic executability; on this basis, complementary curvature constraints between adjacent UAVs are introduced, so that the curvature change of the local flight path segment is inversely related, realizing group streamline cooperation and reducing trajectory aggregation and intersection; the constructed objective function comprehensively considers bending energy, path length, speed and acceleration smoothness and energy consumption, and combines with the normalization and dynamic weight adjustment mechanism to realize adaptive balance of energy utilization and flight path smoothness.

[0008] The UAV cluster cooperative operation path planning method based on distributed computing power according to the embodiment of the present application comprises the following steps:

[0009] According to the obstacles, no-fly zones and terrain boundaries of the operation area, a time-space safety corridor composed of multiple convex polyhedrons is built for each UAV;

[0010] In the time-space safety corridor, an initial connected trajectory is generated according to the take-off and landing points and the task points;

[0011] The initial connected trajectory is reparameterized based on geometric invariants, the flight path is represented as a function of curvature and torsion, and optimization solving is performed in the invariant space;

[0012] The reparameterized trajectory is represented as a curvature-continuous spline curve, the spline satisfies second-order continuity at position, speed and acceleration, and the spline control points and node times are used as optimization variables;

[0013] A target function containing bending energy term and path cost term is constructed, and constraints are applied to the spline curve, wherein complementary curvature constraints are established in the spline parameter matrix corresponding to adjacent UAVs;

[0014] The target function is minimized under the constraints to obtain a curvature-continuous spline flight path that satisfies the complementary curvature constraints;

[0015] When a moving obstacle or a near-collision of adjacent UAVs is detected, the node times of the corresponding corridor sub-section are locally re-optimized to obtain an optimized flight path;

[0016] In the optimization solving and local re-optimization process, the edge computing node is responsible for the synchronization solving and data exchange of the inter-multi-unmanned aerial vehicle constraints, and each unmanned aerial vehicle end independently executes local spline optimization and constraint correction;

[0017] The optimized flight path is issued for execution, and is updated in a sliding time window when the environment changes or the energy threshold is triggered.

[0018] Further, the construction of the space-time safety corridor comprises:

[0019] Obtain the digital elevation model and obstacle boundary information of the work area, and perform spatial discretization processing on the obstacles and no-fly zones;

[0020] Discretize the planning period in the time axis with a preset step size to form a discrete time layer;

[0021] In each discrete time layer, take the current position prediction range of the unmanned aerial vehicle as the center, and construct a spatial convex polyhedron by convex hull generation;

[0022] Based on the unmanned aerial vehicle flight path prediction point set, the spatial density clustering results in adjacent time layers are used to determine the polyhedron vertex coordinates;

[0023] Connect the convex polyhedrons in adjacent time layers in time sequence to form a space-time corridor sequence continuous in the time direction;

[0024] Perform geometric intersection test on the space-time corridor sequence to remove the polyhedrons overlapping with obstacles or no-fly zones;

[0025] Adjust the polyhedron boundary based on path connectivity detection.

[0026] Further, the generation of the initial connected trajectory comprises:

[0027] In the feasible region of the space-time safety corridor, determine the spatial coordinates of the take-off point, the task point and the landing point;

[0028] Use a path search method based on the topology of the feasible corridor to establish a connected relationship graph between the continuous convex polyhedrons;

[0029] The path search method uses parallel heuristic search, maintains the corridor connected relationship graph on the edge node, and synchronizes the search results of each unmanned aerial vehicle;

[0030] Take the take-off point as the starting node and the task point and the landing point as the target node, perform shortest path search in the connected relationship graph to obtain an initial path sequence composed of several convex polyhedron center points;

[0031] The initial path sequence is subjected to line segment splicing and smoothing processing, so that the turning angle, heading rate of change and height rate of change of adjacent line segments all meet the unmanned aerial vehicle dynamics constraints, forming an initial connected trajectory.

[0032] Further, the reparameterization process based on geometric invariants includes:

[0033] A discrete coordinate point sequence of the initial connected trajectory is obtained, and the curvature and torsion of adjacent trajectory segments are calculated.

[0034] The curvature and torsion are used as invariant parameters to construct a parameterized function describing the geometric shape of the trajectory.

[0035] An integral mapping is used to establish a correspondence between the invariant parameters and the spatial coordinates, and the curvature and torsion are used as optimization variables to solve the objective function in the invariant space.

[0036] The integral mapping uses a discrete integral approximation method to realize the mapping of curvature and torsion to spatial coordinates through piecewise summation.

[0037] After solving, the optimization results are mapped back to the spatial coordinate system according to the integral relationship to obtain the corresponding curvature continuous spline trajectory.

[0038] Further, the objective function containing the bending energy term and the path cost term is composed of the weighted sum of the following terms:

[0039] The bending energy term is used to represent the degree of curvature change of the spline curve.

[0040] The path length term is used to represent the path arc length.

[0041] The velocity smoothing term and the acceleration smoothing term are used to constrain the velocity and acceleration rate of change.

[0042] The energy consumption term is used to represent the thrust output or power consumption.

[0043] After normalization, the cost of each item is combined according to the preset weight coefficient to form a dimensionless comprehensive optimization objective.

[0044] The weight coefficients of each cost are dynamically adjusted according to the computing power, power and communication state of the unmanned aerial vehicle.

[0045] Further, the constraints applied to the spline curve include:

[0046] The corridor constraint limits the position of the spline at each discrete time point to be within the space-time safety corridor. The corridor constraint is realized by performing half-space detection on the spline sampling points, and each polyhedron is described by a plane equation set.

[0047] The speed constraint limits the speed of the spline at any node to be less than a certain upper limit.

[0048] acceleration constraints and jerk constraints, which limit the acceleration and jerk of the UAVs within a preset range;

[0049] safety distance constraints, which limit the minimum distance between the UAV and the adjacent UAV to be no less than a safety threshold;

[0050] time window constraints, which limit the arrival time of the UAV at the task point to be within a specified time interval.

[0051] Further, the constraints imposed on the spline curve further include complementary curvature constraints, and the establishing process comprises:

[0052] obtaining the spline parameter matrix of the adjacent UAVs within the same time period, and discretely representing the local curvature distribution of each spline;

[0053] establishing a complementary relationship constraint between the discretized curvature distributions, so that the curvature of the corresponding adjacent UAV decreases by a preset inverse ratio coefficient when the local curvature of any UAV increases on the flight path segment;

[0054] The inverse ratio coefficient is dynamically set according to the distance between the UAVs and the communication delay, and the complementary curvature data is synchronously updated by the edge node.

[0055] In the spline optimization solving process, the complementary relationship is incorporated into the objective function solving in the form of linear or nonlinear constraints.

[0056] Further, the process of minimizing the objective function under the constraints and performing local re-optimization comprises:

[0057] The spline control points and node times are jointly solved in an iterative optimization manner to obtain a curvature continuous spline flight path that satisfies the constraint conditions.

[0058] The local re-optimization is performed within a preset time window, and the optimization range is limited to the detected corridor sub-segment and its adjacent time layer.

[0059] When a moving obstacle or a near conflict with an adjacent UAV is detected, the corresponding corridor sub-segment is determined, the node time of the sub-segment is taken as a local optimization variable, the objective function is re-solved under the condition of maintaining the continuity of adjacent sub-segments, and an optimized flight path is obtained.

[0060] Further, the step of issuing the optimized flight path for execution comprises:

[0061] The optimized flight path parameters corresponding to each UAV are issued to the onboard control unit in segments, and the flight path tracking control is performed in time sequence.

[0062] Upon detecting a change in environmental parameters or energy threshold triggering, re-acquire state information and invoke the aforementioned optimization steps for rolling update within a sliding time window to generate new feasible trajectories and replace the original ones.

[0063] Further, the rolling update within the sliding time window includes:

[0064] Real-time monitoring of the pose state, remaining energy, communication link quality and external environmental parameters of the UAV;

[0065] When any parameter is detected to exceed the preset threshold, trigger the rolling update process;

[0066] In the rolling update process, the latest optimization result is taken as the initial condition to re-execute spline optimization and constraint correction for the affected corridor sub-sections;

[0067] In the rolling update process, the edge computing node is responsible for multi-UAV state aggregation and unified constraint synchronization, and each UAV end executes local optimization and returns the update result;

[0068] The updated trajectory is distributed to the corresponding UAV through the edge computing node.

[0069] The beneficial effects of the present application are:

[0070] The present application divides the trajectory solving process into two levels of edge node and airborne end, the edge computing node is responsible for the synchronous solving and data exchange of multi-UAV constraints, and each UAV end independently executes local spline optimization and constraint correction, forming a parallel computing structure, which effectively reduces the load of the central node and improves the real-time performance and scalability of the path planning.

[0071] By converting the trajectory from spatial coordinate representation to invariant representation of curvature and torsion, optimization solving is performed in the invariant space, which can significantly reduce parameter correlation, make the objective function have better convergence characteristics, and make the optimization result more stable in terms of curvature continuity and dynamics executability.

[0072] The curvature change of adjacent UAVs in the local trajectory segment is inversely related, when one UAV performs a large steering operation, the adjacent UAV automatically adjusts its trajectory curvature, realizes the streamline cooperation between the groups, reduces the risk of trajectory aggregation and intersection, and improves the safety and space utilization of the cluster flight.

[0073] The objective function considers factors such as bending energy, path length, speed and acceleration smoothness, and energy consumption, and through normalization and dynamic weight adjustment mechanism, it realizes the trade-off between energy and smoothness under different flight states, and enhances the adaptability of the planning result. BRIEF DESCRIPTION OF DRAWINGS

[0074] The accompanying drawings are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and are meant to explain the application without limiting the application to the exact embodiments shown. In the drawings:

[0075] Figure 1 A flow chart of the method for planning a path for cooperative operation of a UAV cluster based on distributed computing power according to the present application;

[0076] Figure 2 A schematic diagram of the principle of constructing a time-space safety corridor for the method for planning a path for cooperative operation of a UAV cluster based on distributed computing power according to the present application;

[0077] Figure 3 A schematic diagram of local re-optimization and rolling update of a sliding time window for the method for planning a path for cooperative operation of a UAV cluster based on distributed computing power according to the present application. DETAILED DESCRIPTION

[0078] The application will now be described in further detail with reference to the drawings. These drawings show only the essential features of the application and are therefore to be regarded only as a schematic representation of the basic structure of the application.

[0079] Reference Figure 1 - Figure 3 The method for planning a path for cooperative operation of a UAV cluster based on distributed computing power comprises the following steps:

[0080] According to the obstacles, no-fly zones and terrain boundaries of the operation area, a time-space safety corridor composed of multiple convex polyhedrons is constructed for each UAV;

[0081] Within the time-space safety corridor, an initial connected trajectory is generated according to the take-off and landing points and the task points;

[0082] The initial connected trajectory is re-parameterized based on geometric invariants, the flight path is represented as a function of curvature and torsion, and optimization is performed in the invariant space;

[0083] The re-parameterized trajectory is represented as a spline curve with continuous curvature, and the spline satisfies second-order continuity at position, velocity and acceleration, and the spline control points and node times are used as optimization variables;

[0084] A target function containing a bending energy term and a path cost term is constructed, and constraints are imposed on the spline curve, wherein complementary curvature constraints are established in the spline parameter matrices corresponding to adjacent UAVs;

[0085] The target function is minimized under the constraints to obtain a curvature-continuous spline flight path that satisfies the complementary curvature constraints;

[0086] When a moving obstacle or a near collision of adjacent UAVs is detected, the node time of the corresponding corridor sub-section is locally re-optimized to obtain an optimized flight path;

[0087] In the optimization solving and local re-optimization process, the edge computing node is responsible for the synchronous solving and data exchange of the inter-UAV constraints, and each UAV end independently executes local spline optimization and constraint correction;

[0088] The optimized flight path is issued for execution, and is rolled over and updated in a sliding time window when the environment changes or the energy threshold is triggered.

[0089] In the embodiment, the construction of the spatiotemporal safety corridor includes:

[0090] The digital elevation model and obstacle boundary information of the work area are obtained, the resolution of the digital elevation model is set to 1 meter, a regular grid structure is used, the obstacle boundary information is generated based on laser radar point cloud and geographic information database, and after three-dimensional gridding processing, it is unified to the global coordinate system, the spatial discretization processing is performed on the obstacles and no-fly zones, the continuous region is divided into voxel units, and the voxel edge length is taken as 2 meters;

[0091] The planning period is discretized on the time axis at a preset step length to form a discrete time layer, and each time layer represents a static space snapshot in a fixed time interval;

[0092] In each discrete time layer, the spatial position of the UAV at the future time is predicted according to the spatial position, velocity, acceleration and attitude angle of the UAV at the current time, taking the current position of the UAV as the center, the predicted position is calculated by adding the product of the velocity and the time step to the current position, and then adding half of the product of the acceleration and the square of the time step, based on these predicted positions, a predicted point set containing multiple sampling points is generated;

[0093] According to the obtained predicted point set, a spatial convex polyhedron is constructed by convex hull generation, taking the predicted point set as input, the smallest convex hull that can completely envelope all predicted points is calculated, so as to determine the vertex coordinates and boundary plane parameters of the polyhedron, and each convex polyhedron generated represents the safe flight space of the UAV in the time layer;

[0094] Based on the UAV flight path prediction point set, the spatial density clustering result in the adjacent time layer is used to determine the polyhedron vertex coordinates, the predicted points are projected onto the three-dimensional grid, the point density in each grid cell is counted, and the points with a density not less than 80% of the maximum density are selected as the polyhedron vertex candidate points, the candidate points of adjacent time layers are matched according to the spatial nearest neighbor principle, and the corresponding vertex of the convex polyhedron in the continuous time layer is determined;

[0095] The convex polyhedrons in adjacent time layers are connected in time sequence to form a time-space corridor sequence continuous in time direction, geometric intersection test is performed on the time-space corridor sequence, polyhedrons intersecting with obstacles or no-fly zones are removed, and the boundary equation of each convex polyhedron is substituted into the grid point coordinates of the obstacle, and if more than 5% of the grid points are found to be located inside the polyhedron, it is determined that the polyhedron intersects with the obstacle;

[0096] The polyhedron boundary is adjusted based on path connectivity detection.

[0097] In the embodiment, the generation of the initial connected trajectory includes:

[0098] In the feasible region of the time-space safe corridor, the spatial coordinates of the takeoff point, the task point and the landing point are determined, the takeoff point and the landing point are determined according to the spatial deployment of the operation task, and the task point is set according to the operation target or the aerial survey coverage;

[0099] A path search method based on the topology of the feasible corridor is adopted to establish a connected relationship graph between the continuous convex polyhedrons, each convex polyhedron is regarded as a node in the graph structure, and the polyhedrons having spatial intersection between adjacent time layers are regarded as connectable nodes.

[0100] The path search method adopts parallel heuristic search, maintains the corridor connected relationship graph on the edge nodes and synchronizes the search results of each unmanned aerial vehicle, the path search starts from the polyhedron node where the takeoff point is located, takes the polyhedrons where the task point and the landing point are located as target nodes, and searches the feasible path layer by layer along the connected edges of the nodes, in the search process, the optimal channel is selected according to the path cost function, and the cost function considers the spatial distance, the change of steering and the connectivity index;

[0101] Line segment splicing and smoothing processing are performed on the initial path sequence, in the splicing process, the center points of adjacent polyhedrons are connected to form an initial flight path line segment in time sequence, the turning angle, the change rate of heading and the change rate of height of adjacent line segments all meet the unmanned aerial vehicle dynamics constraints, and an initial connected trajectory is formed.

[0102] In the embodiment, the reparameterization process based on geometric invariants includes:

[0103] A discrete coordinate point sequence of the initial connected trajectory is obtained, the curvature and the torsion of adjacent trajectory segments are calculated, the curvature is used to represent the bending degree of the trajectory at the position, and the torsion is used to represent the spatial torsion degree of the trajectory;

[0104] The initial connected trajectory is converted from the coordinate sequence form to a function representation form with curvature and torsion as independent variables by taking the curvature and the torsion as invariant parameters;

[0105] The invariant parameters are corresponded to the spatial coordinates by integral mapping, and the objective function is solved in the invariant space with curvature and torsion as optimization variables. The optimization process is constrained by curvature continuity and torsion smoothness, and the change rates of curvature and torsion are iteratively adjusted to meet the requirements of smoothness and stability in a continuous time period, so as to reduce the overall bending energy and spatial deviation.

[0106] The integral mapping adopts a discrete integral approximation method to realize the mapping of curvature and torsion to spatial coordinates by piecewise summation.

[0107] After the solution is completed, the optimization results are mapped back to the spatial coordinate system according to the integral relationship to obtain the corresponding curvature continuous spline trajectory.

[0108] After the reparameterization based on curvature and torsion is completed, the curvature continuous spline optimization and constraint application process is performed. The reparameterized trajectory is expressed in the form of a spline curve. The spline curve is defined by a plurality of control points and node times, and the spatial coordinates of each control point and the time nodes constitute the optimization variables.

[0109] The objective function is composed of the weighted combination of a plurality of cost items, including a bending energy item for representing the geometric smoothness of the trajectory, a path cost item for measuring the path length, a speed and acceleration smoothness item for reflecting the motion smoothness, and an energy cost item for controlling the energy use efficiency.

[0110] ;

[0111] denotes the curvature of the trajectory at the arc length parameter denotes the torsion denotes the derivative of the velocity vector with respect to the arc length, which is used to describe the rate of change of the velocity, denotes the derivative of the acceleration vector with respect to the arc length, which is used to describe the acceleration smoothness, denotes the total arc length of the trajectory, 、 、 、 、 、 is a weighting coefficient, which is normalized and adjusted according to the different requirements of the task for smoothness and energy, 、 、 are respectively the normalized reference length, the velocity and the acceleration constant, which are used to keep each item dimensionless.

[0112] In the embodiment, the constraints applied to the spline curve include:

[0113] Corridor constraint, limiting the position of the spline at each discrete time point to be within a safe corridor in space-time, the corridor constraint is achieved by performing half-space detection on the sample points of the spline, and each polyhedron is described by a set of plane equations;

[0114] Velocity constraint, limiting the velocity of the spline at any node to be no more than a set upper limit;

[0115] Acceleration constraint and jerk constraint, limiting the acceleration and jerk of the UAV within a preset range;

[0116] Safety distance constraint, limiting the minimum distance between the UAV and the adjacent UAV to be no less than a safety threshold;

[0117] Time window constraint, limiting the arrival time of the UAV at the task point to be within a specified time interval.

[0118] In the embodiment, the constraints imposed on the spline curve also include a complementary curvature constraint, and the establishment process includes:

[0119] Obtaining the spline parameter matrix of the adjacent UAV within the same time period, and discretely representing the local curvature distribution of each spline;

[0120] Establishing a complementary relationship constraint between the discretized curvature distributions, so that the curvature of the corresponding adjacent UAV decreases by a preset inverse ratio coefficient when the local curvature of any UAV increases;

[0121] The inverse ratio coefficient is dynamically set according to the distance between the UAVs and the communication delay, and the complementary curvature data is synchronized and updated by the edge node;

[0122] In the spline optimization solving process, the complementary relationship is incorporated into the objective function solving in the form of linear or nonlinear constraints.

[0123] After all the constraints are imposed, the objective function is minimized. In the solving process, the spline control points and node times are adjusted iteratively to make the objective function value converge to the optimal value while satisfying all the constraints. After the optimization is completed, the spline trajectory with continuous curvature, smooth torsion and meeting the requirements of multi-UAV cooperation is obtained.

[0124] The consistency of the optimization result is checked. By comparing the change curves of the curvature and the torsion before and after optimization, it is confirmed that the continuity and smoothness meet the preset standard.

[0125] In the embodiment, the process of minimizing the objective function under the constraints and locally re-optimizing includes:

[0126] The spline control points and node times are jointly solved by using an iterative optimization method to obtain a curvature-continuous spline trajectory that satisfies the constraint conditions;

[0127] The local re-optimization is performed within a preset time window, and the optimization range is limited to the detected corridor sub-section and its adjacent time layer;

[0128] When a moving obstacle or a near conflict with a neighboring UAV is detected, the corresponding corridor sub-section is determined, the node time of the sub-section is taken as a local optimization variable, the objective function is re-solved under the condition of maintaining the continuity of adjacent sub-sections, and an optimized flight path is obtained.

[0129] In the embodiment, the step of issuing the optimized flight path for execution includes:

[0130] The optimized flight path parameters corresponding to each UAV are segmented and issued to the onboard control unit, and the flight path tracking control is performed in time sequence;

[0131] When an environmental parameter change or an energy threshold is detected, the state information is re-collected, and the aforementioned optimization step is called within a sliding time window for rolling update to generate a new feasible flight path and replace the original flight path.

[0132] In the embodiment, the rolling update within the sliding time window includes:

[0133] The UAV continuously collects its own position, speed, remaining energy, communication link state and external obstacle information during flight;

[0134] When any parameter is detected to exceed a preset threshold, the rolling update process is triggered, and the range of local re-optimization is limited to the conflict corridor sub-section and its adjacent time layer;

[0135] In the local re-optimization process, the spline parameters of the non-affected area are kept unchanged, only the spline control points and node times in the conflict area are taken as optimization variables, the optimization objective is the same as in global planning, and the objective function value is still minimized based on the curvature and torsion smoothing principle, but the boundary conditions of the optimization area are fixed as the endpoints of the adjacent trajectory sections;

[0136] The edge computing node is responsible for synchronous constraint update and data distribution after receiving the local state of each UAV. If a conflict is detected between multiple UAVs in the same area, the edge computing node coordinates the complementary curvature constraint relationship of each UAV through a unified constraint synchronization mechanism.

[0137] The updated flight path is distributed to the corresponding UAV through the edge computing node.

[0138] During long-term flight, a sliding time window mechanism is adopted to cope with the continuous changes in the environment and energy consumption differences. The length of the time window is dynamically set according to the task period and communication frequency. The system reacquires the flight state, obstacle information and energy level at the start of each time window, and re-executes the spline optimization and constraint correction process with the latest optimization result as the initial condition.

[0139] When communication delay or data loss is detected, the edge computing node determines the validity of each UAV state according to the latest synchronization timestamp. If data lag occurs, the last optimization result is kept unchanged and the update is delayed to the next window period to ensure trajectory continuity and system stability.

[0140] Embodiment 1:

[0141] To verify the feasibility of the present application, the distributed computing-based UAV cluster cooperative operation path planning method of the present application is applied to the mountain power inspection task. In this embodiment, the operation area contains complex terrain, high difference and multi-obstacle environment, and a number of UAVs equipped with edge computing units form a cluster. Each UAV can independently perform spline optimization and constraint correction, and the edge computing node is responsible for global constraint synchronization and rolling update management.

[0142] According to the regional digital elevation model and obstacle boundary information, a spatial grid of the operation area is discretely generated. The edge computing node performs discretization on the time axis to form a multi-layer spatio-temporal safety corridor. Through the convex hull generation method, the feasible flight channel of each UAV is established, and geometric intersection detection and connectivity detection are performed.

[0143] Taking the take-off point and inspection task point of each UAV as input, an initial connected trajectory is constructed. After smoothing, the trajectory is input to the UAV end to perform invariant parameterization and spline optimization based on curvature and torsion. The edge computing node synchronizes the complementary curvature constraints of each UAV during this process, and dynamically adjusts the constraint coefficients according to the distance and communication delay.

[0144] When a moving obstacle or potential conflict with adjacent UAVs is detected, local re-optimization is triggered. The affected UAVs only adjust the spline control points and node times for the corresponding corridor segments, and the edge node is responsible for conflict segment synchronization solution and trajectory coordination.

[0145] During the entire inspection, the edge computing node periodically performs rolling update in a sliding time window manner to correct the flight path in real time. After each update period, the multi-UAV state information is reassembled and constraint correction is performed to achieve continuous adaptation of the flight path and maintenance of group stability.

[0146] To verify the performance of the application, the method of the application and the traditional centralized path planning method are compared under the same working conditions. The experiment adopts six multi-rotor unmanned aerial vehicles to form a cluster and performs inspection and obstacle avoidance tasks in a power line inspection area about 3 kilometers long. The experimental environment includes multiple obstacles, terrain undulations and communication delay interference.

[0147] Table 1 Comparison of experimental parameter settings

[0148] Parameter Category Invention Method (Distributed) Control Method (Centralized) Number of UAVs 6 6 Length of Operation Area About 3 km About 3 km Number of Edge Nodes 1 1 central control server Communication Delay About 120 ms About 120 ms Optimized Refresh Cycle 2 s (dynamic) 10 s (fixed) Trajectory Representation Form Curvature-flexibility spline Three-dimensional coordinate spline Optimization Range Local window re-optimization Global single solution Complementary constraint mechanism Enabled Not enabled

[0149] Under the same hardware conditions, the experimental results of the two methods in terms of calculation time, trajectory smoothness, communication delay adaptability, energy utilization rate, and cluster safety distance retention rate are respectively counted.

[0150] Table 2 Comparison of performance of the method of the application and the control method

[0151] Index Unit Control Method (Centralized) Invention Method (Distributed) Improvement Amplitude Average path planning time s 4.86 1.72 ↓64.6% Curvature continuity error (mean square error) 1 / m² 0.031 0.009 ↓70.9% Control stability rate under communication delay % 82.3 97.5 ↑18.5% Cluster safety distance maintenance rate % 91.4 99.2 ↑8.5% Energy utilization rate % 84.7 94.1 ↑11.1% Central node CPU occupancy rate % 92.8 28.3 ↓69.5%

[0152] As can be seen from Table 2, the method of the application reduces the average path planning time by about 65% compared to the traditional method, reduces the trajectory curvature continuity error by about 71%, improves the flight control stability rate under communication delay conditions by nearly 20%, maintains the cluster safety distance retention rate close to 100%, and significantly reduces the center node load. This shows that the application realizes the comprehensive improvement of the real-time performance, smoothness and multi-unmanned aerial vehicle cooperation performance of path planning by combining edge computing with geometric invariant optimization.

[0153] To ensure objectivity, the experiment is repeated ten times, and the data in the table are averaged. All results are only used to illustrate the technical effects of the application and do not limit the protection scope of the application.

[0154] The above is only the preferred specific implementation of the application, but the protection scope of the application is not limited thereto. Any person skilled in the art can make equivalent replacements or changes to the technical solutions and inventive concepts of the application within the technical scope disclosed by the application, which should be covered within the protection scope of the application.

Claims

1. A method for path planning of cooperative work of a UAV cluster based on distributed computing power, characterized in that, The method comprises the following steps: According to the obstacles, no-fly zones and terrain boundaries of the work area, a time-space safety corridor composed of multiple convex polyhedrons is established for each unmanned aerial vehicle; Within the time-space safety corridor, an initial connected trajectory is generated according to the take-off point and the task point; The initial connected trajectory is re-parameterized based on geometric invariants, the flight path is represented as a function of curvature and torsion, and optimization is performed in the invariant space; The re-parameterized trajectory is represented as a curvature-continuous spline curve, and the spline satisfies the second-order continuity at the positions, velocities and accelerations, and the spline control points and node times are used as optimization variables; A target function containing a bending energy term and a path cost term is constructed, and constraints are imposed on the spline curve, wherein complementary curvature constraints are established in the spline parameter matrices corresponding to adjacent unmanned aerial vehicles; The target function is minimized under the constraints to obtain a curvature-continuous spline trajectory that satisfies the complementary curvature constraints; When a moving obstacle or a near-collision between adjacent unmanned aerial vehicles is detected, the node times of the corresponding corridor sub-segments are locally re-optimized to obtain an optimized trajectory; During the optimization and local re-optimization processes, the edge computing nodes are responsible for the synchronous solving and data exchange of the constraints between multiple unmanned aerial vehicles, and each unmanned aerial vehicle independently performs local spline optimization and constraint correction; The optimized trajectory is executed and is updated in a sliding time window when the environment changes or the energy threshold is triggered.

2. The method of claim 1, wherein, The construction of the time-space safety corridor comprises: Obtain the digital elevation model and obstacle boundary information of the work area, and perform spatial discretization processing on the obstacles and no-fly zones; Discretize the planning period on the time axis at a preset step length to form discrete time layers; In each discrete time layer, a spatial convex polyhedron is constructed by convex hull generation with the current position prediction range of the unmanned aerial vehicle as the center; Based on the unmanned aerial vehicle trajectory prediction point set, the spatial density clustering results in adjacent time layers are used to determine the polyhedron vertex coordinates; The convex polyhedrons in adjacent time layers are connected in time sequence to form a time-space corridor sequence that is continuous along the time direction; Perform geometric intersection test on the time-space corridor sequence to remove the polyhedrons that overlap with obstacles or no-fly zones; Adjust the polyhedron boundary based on path connectivity detection.

3. The method of claim 1, wherein, The generation of the initial connected trajectory comprises: Determine the spatial coordinates of the take-off point, the task point and the landing point in the feasible region of the time-space safety corridor; Use a path search method based on the topology of the feasible corridor to establish a connected relationship graph between the continuous convex polyhedrons; The path search method uses parallel heuristic search, maintains the corridor connectivity relationship graph on the edge node, and synchronizes the search results of each unmanned aerial vehicle; Perform shortest path search in the connected relationship graph with the take-off point as the starting node and the task point and the landing point as the target nodes to obtain an initial path sequence composed of several convex polyhedron center points; Perform line segment splicing and smoothing processing on the initial path sequence to make the turning angles, heading rate and height rate of adjacent line segments meet the unmanned aerial vehicle dynamics constraints, and form an initial connected trajectory.

4. The method of claim 1, wherein, The re-parameterization process based on geometric invariants comprises: Obtain the discrete coordinate point sequence of the initial connected trajectory, and calculate the curvature and torsion of adjacent trajectory segments; The curvature and the torsion are taken as invariant parameters to construct a parameterized function describing the geometric shape of the flight path; An integral mapping is used to establish a correspondence between the invariant parameters and the spatial coordinates, and the curvature and the torsion are taken as optimization variables to solve the objective function in the invariant space; The integral mapping adopts a discrete integral approximation method to realize the mapping of the curvature and the torsion to the spatial coordinates through piecewise summation; After the solution is completed, the optimization results are mapped back to the spatial coordinate system according to the integral relationship to obtain the corresponding curvature continuous spline trajectory.

5. The method of claim 1, wherein, The objective function comprising the bending energy term and the path cost term is composed of a weighted sum of the following terms: The bending energy term is used to represent the degree of curvature change of the spline curve; The path length term is used to represent the path arc length; The velocity smoothing term and the acceleration smoothing term are used to constrain the velocity and acceleration change rate; The energy consumption term is used to represent the thrust output or the power consumption; After normalization, each cost is combined according to the preset weight coefficient to form a dimensionless comprehensive optimization objective; The weight coefficients of each cost are dynamically adjusted according to the computing power, power and communication state of the unmanned aerial vehicle.

6. The method of claim 1, wherein, The constraints imposed on the spline curve include: The corridor constraint limits the positions of the spline at each discrete time point to be located within the space-time safety corridor, and the corridor constraint is realized by performing half-space detection on the spline sampling points, each polyhedron being described by a plane equation set; The velocity constraint limits the velocity of the spline at any node to be not greater than a set upper limit; The acceleration constraint and the jerk constraint limit the acceleration and the jerk of the unmanned aerial vehicle within a preset range; The safety distance constraint limits the minimum distance between the unmanned aerial vehicle and the adjacent unmanned aerial vehicle to be not less than a safety threshold; The time window constraint limits the arrival time of the unmanned aerial vehicle at the task point to be within a specified time interval.

7. The method of claim 1, wherein, The constraints imposed on the spline curve also include a complementary curvature constraint, and the establishment process includes: Obtaining the spline parameter matrix of the adjacent unmanned aerial vehicle within the same time period to discretely represent the local curvature distribution of the respective spline; Establishing a complementary relationship constraint between the discretized curvature distributions, so that when the curvature of any unmanned aerial vehicle increases at a local path segment, the curvature of the corresponding adjacent unmanned aerial vehicle decreases by a preset inverse ratio coefficient; The inverse ratio coefficient is dynamically set according to the distance between the unmanned aerial vehicles and the communication delay, and the complementary curvature data is synchronized by the edge node; During the spline optimization solving process, the complementary relationship is incorporated into the objective function solving in the form of linear or nonlinear constraints.

8. The method of claim 1, wherein, The process of minimizing the objective function under the constraints and performing local re-optimization includes: The iterative optimization method is used to jointly solve the spline control points and the node time to obtain a curvature continuous spline trajectory that satisfies the constraint conditions; The local re-optimization is performed within a preset time window, and the optimization range is limited to the detected corridor sub-section and its adjacent one layer of time layer; When a moving obstacle or a near conflict with the adjacent unmanned aerial vehicle is detected, the corresponding corridor sub-section is determined, the node time of the sub-section is taken as a local optimization variable, and the objective function is re-solved under the condition of maintaining the continuity of the adjacent sub-section to obtain an optimized flight path.

9. The method of claim 1, wherein, The step of executing the optimized flight path includes: The optimal flight path parameters of each UAV are segmented and sent to the on-board control unit, and the flight path tracking control is executed in time sequence; When the environmental parameter changes or the energy threshold is triggered, the state information is re-acquired, and the aforementioned optimization steps are called within the sliding time window for rolling update to generate a new feasible flight path and replace the original flight path.

10. The method of claim 1, wherein, The rolling update within the sliding time window includes: The pose state, remaining energy, communication link quality and external environmental parameters of the UAV are monitored in real time; When any parameter is detected to be beyond the preset threshold, the rolling update process is triggered; In the rolling update process, the spline optimization and constraint correction are re-executed for the affected corridor sub-sections with the latest optimization result as the initial condition; In the rolling update process, the edge computing node is responsible for the state aggregation and unified constraint synchronization of multiple UAVs, and each UAV end executes local optimization and returns the update result; The updated flight path is distributed to the corresponding UAV through the edge computing node.

Images