A method for cooperative delivery path planning of a UAV and an unmanned vehicle

Abstract

This invention discloses a method for collaborative delivery path planning using unmanned aerial vehicles (UAVs) and unmanned vehicles (UAVs), specifically relating to the field of unmanned delivery path planning technology. It establishes a unified spatiotemporal evolution model by constructing a cross-domain environmental parameter set and a cross-coupling relationship graph, and introduces temporal phase synchronization constraints to achieve unified modeling of air-to-ground collaborative paths. Furthermore, it constructs a mismatch propagation function to quantify the impact of UAV trajectory deviation on the UAV path, and calculates the path-level mismatch gain coefficient by combining the spectral radius index and the total graph variation. Simultaneously, it calculates the path phase entropy value based on the information entropy method to identify mismatch accumulation segments. Based on this, it introduces path recoverability coding to redundantly reconstruct unstable regions, and iteratively optimizes the objective function through joint optimization, ultimately obtaining a collaborative delivery path scheme that meets energy consumption constraints and possesses temporal convergence characteristics. This invention can effectively suppress mismatch propagation and improve path stability and collaborative delivery reliability.

Description

Technical Field

[0001] This invention relates to the field of unmanned delivery route planning technology, specifically to a method for collaborative delivery route planning using drones and unmanned vehicles. Background Technology

[0002] In areas such as canyon wind corridors, gaps between super high-rise buildings, and areas where communication is damaged after disasters, the collaborative delivery of drones and unmanned vehicles is subject to not only dynamic access restrictions and energy consumption constraints, but also a rare but high-risk "cross-domain temporal mismatch cumulative effect". That is, the hovering deviation of drones caused by transient airflow disturbances will be transmitted to the path decision of unmanned vehicles through the handover node, resulting in repeated local path corrections and the formation of non-convergent collaborative oscillations. This problem is particularly prominent when multiple tasks are concurrent and communication is intermittently interrupted. Existing methods do not consider the cross-carrier propagation mechanism of mismatch, making it difficult to achieve stable collaboration. Summary of the Invention

[0003] The purpose of this invention is to provide a method for collaborative delivery route planning using drones and unmanned vehicles, in order to address the shortcomings of the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for collaborative delivery route planning using unmanned aerial vehicles (UAVs) and unmanned vehicles, comprising: Acquire the time-varying micro-airflow field, dynamic constraint area, and communication accessibility parameters of the delivery area to construct a cross-domain environmental parameter set E; Based on the delivery task and vehicle status, establish a handover and coupling relationship diagram G between drones and unmanned vehicles; A unified spatiotemporal evolution model W is constructed based on E and G, and a temporal phase synchronization constraint is introduced to ensure that the UAV flight trajectory and the UAV driving trajectory satisfy the phase alignment relationship at the handover node. In W, a mismatch propagation function F is constructed to map the trajectory offset generated by the UAV to the path disturbance response of the unmanned vehicle, thus obtaining the initial cooperative path set P; The stability of P is determined, the mismatch accumulation section caused by F is identified, and the set of unstable regions Q is generated. In the region corresponding to Q, path recoverability coding is introduced to perform redundant segmentation and reconstruction marking on the path, and W is locally reconstructed based on the coding to obtain the corrected model W′. Based on W′, iterative optimization is performed to ensure that the path meets energy consumption constraints while suppressing mismatch propagation and achieving cross-carrier temporal convergence, outputting the final collaborative delivery path scheme.

[0005] Preferably, establishing the handover and coupling relationship diagram G between the UAV and the unmanned vehicle includes the following steps: Based on the handover requirements and vehicle operation status in the delivery task, a set of candidate handover nodes for drones and unmanned vehicles is extracted, and the candidate handover nodes are matched and filtered based on spatial location and time window to form an initial handover node pair; For the initial handover node pair, based on the vehicle's remaining energy, load capacity, and operational constraints, a feasible handover constraint relationship between the nodes is constructed, and weighted connection edges are generated accordingly. Based on the weighted connection edges, a temporal consistency constraint is introduced to perform alignment calculations on the arrival time of the UAV and the arrival time of the unmanned vehicle, and to determine the set of valid edges that meet the synchronous handover conditions. Based on the set of effective edges and the corresponding intersection nodes, an intersection coupling relationship graph with spatiotemporal coupling properties is constructed.

[0006] Preferably, the construction of a unified spatiotemporal evolution model W based on E and G includes the following steps: The spatial grid and time series information in the cross-domain environmental parameter set are mapped to the nodes and connecting edges in the intersection and coupling relationship graph, an extended spatiotemporal node set containing the time dimension is constructed, and each node is assigned environmental constraint attributes and time markers. Based on the extended spatiotemporal node set, a UAV flight trajectory function and an unmanned vehicle driving trajectory function are established, and the trajectories are sampled by discrete time steps to form corresponding trajectory sequences. A temporal phase function is constructed based on the trajectory sequence. The temporal phase function is used to characterize the difference in arrival stages between the UAV and the unmanned vehicle at the same handover node, and a phase difference threshold is defined to constrain the degree of synchronization between the two. The trajectory sequences are adjusted and filtered according to the phase difference threshold, and the trajectory combinations that satisfy the phase alignment relationship are retained, thereby forming a unified spatiotemporal evolution model.

[0007] Preferably, the step of mapping the trajectory offset generated by the UAV to the path disturbance response of the unmanned vehicle to obtain the initial cooperative path set P includes the following steps: The trajectory offset of the UAV is extracted at the junction nodes in the unified spatiotemporal evolution model, and the continuous offset change rate is calculated based on the time series to form the UAV side offset feature sequence. A mismatch propagation function is constructed based on the offset feature sequence. The mismatch propagation function takes the offset change rate and the time interval between nodes as input and calculates the path disturbance response of the UAV through weighted mapping. The path disturbance response is superimposed on the original driving trajectory of the UAV to correct the disturbance of the UAV path and obtain the disturbed path set. The disturbed UAV path is coupled and matched with the UAV flight trajectory to select the path combination that meets the temporal phase synchronization constraint, thereby generating the initial cooperative path set.

[0008] Preferably, the stability determination of P includes the following steps: For each handover node in the initial cooperative path set, calculate the path-level mismatch gain coefficient; Based on the temporal phase difference data of each handover node, a probability distribution is constructed according to the preset interval division rules, and the path phase entropy value is obtained through the information entropy calculation method. The path-level mismatch gain coefficient and the path phase entropy value are jointly mapped to construct a two-dimensional stability determination space, and each path segment is classified and labeled according to a preset stability boundary function. Path segments that are deemed not to meet stability boundary conditions are identified as mismatch accumulation segments and aggregated to form a set of unstable regions.

[0009] Preferably, the calculation of the path-level mismatch gain coefficient includes the following steps: Based on the connection relationship between each handover node in the initial cooperative path set, a mismatch transfer matrix is ​​constructed. The matrix elements of the mismatch transfer matrix are determined by the ratio of the unmanned vehicle path disturbance response between adjacent nodes to the UAV trajectory offset. The mismatch transfer matrix is ​​decomposed into eigenvalues, and the magnitude of its largest eigenvalue is calculated as the spectral radius index. The spectral radius index is then used as a characterization parameter of the overall mismatch amplification capability of the path. Map the mismatch corresponding to each junction node to the node attributes of the junction coupling relationship graph to construct the graph signal, and calculate the total graph variation value based on the connection edges between nodes; The path-level mismatch gain coefficient is obtained by weighted fusion calculation of the spectral radius index and the total variation value of the graph.

[0010] Preferably, the calculation of the path phase entropy value includes the following steps: The timing phase difference data of each handover node in the initial collaborative path set is obtained, and the timing phase difference is segmented according to the preset interval division rule. The preset interval is divided into equal or non-equal intervals according to the phase difference value range to form multiple phase intervals. The number of nodes in each phase interval is counted, and the probability distribution value of each interval is calculated by the proportion of the number of nodes to the total number of nodes, thereby constructing the phase probability distribution sequence of the path; Based on the phase probability distribution sequence, the probability values ​​of each interval are weighted and summed according to the information entropy calculation formula to obtain the path phase entropy value.

[0011] Preferably, the step of introducing path recoverability coding in the Q-corresponding region and performing redundant segmentation and reconstruction marking on the path includes the following steps: For each path segment in the set of unstable regions, a path recoverability coding sequence is constructed. The path recoverability coding sequence consists of a path segment identifier, a mismatch level identifier, and an alternative path index to characterize the reconstruction priority and replacement relationship of each segment. Based on the path recoverability coding sequence, the unstable path segment is subjected to redundant segmentation processing. At least one alternative path is generated in each segment, and a set of redundant paths that meet the energy consumption and timing requirements is selected through constraints. Based on the substitution relationship in the path recoverability coding sequence, the original path segment is reconstructed and marked, and the selected redundant path is embedded into the original path structure to form an updated path connection relationship. Based on the updated path connection relationship, the corresponding nodes and connecting edges in the unified spatiotemporal evolution model are locally replaced and reconstructed to obtain the corrected unified spatiotemporal evolution model.

[0012] Preferably, the iterative optimization based on W′ includes the following steps: In the modified unified spatiotemporal evolution model, a joint optimization objective function is constructed. The joint optimization objective function consists of an energy consumption term, a mismatch propagation suppression term, and a time phase convergence term. Each term is linearly combined through weighted coefficients to form a comprehensive evaluation index. Based on the joint optimization objective function, the path is iteratively updated and calculated. A combination of gradient descent and path rearrangement is used to gradually adjust the node connection order and time label. In each iteration, the energy consumption, mismatch propagation intensity and phase difference change of the path are calculated and compared with the preset energy consumption threshold, mismatch threshold and phase convergence threshold. When any constraint is not satisfied, the corresponding path segment is locally corrected. When the results of multiple consecutive iterations meet the convergence criteria, the iteration process stops, and the final collaborative delivery path scheme that satisfies energy consumption constraints and achieves cross-carrier temporal convergence is output.

[0013] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. This invention constructs a cross-domain environmental parameter set, a cross-coupling relationship graph, and a unified spatiotemporal evolution model. By introducing a mismatch propagation function and temporal phase synchronization constraints during path planning, it achieves deep coupling optimization of UAVs and unmanned vehicles in complex dynamic environments. Compared to existing methods based solely on static paths or individual optimization, this invention can characterize disturbance propagation relationships in advance during the path generation stage and quantify the mismatch amplification trend through spectral radius indices and total graph variation values. This effectively identifies potentially unstable path segments, avoiding path oscillations, task delays, or energy consumption runaway during collaborative delivery, significantly improving the stability of path planning.

[0014] 2. This invention introduces path recoverability coding and a local reconstruction strategy to generate and dynamically replace redundant paths within unstable regions. It then combines this with a joint optimization objective function for iterative convergence calculation, enabling paths to meet energy constraints while suppressing mismatch propagation and achieving temporal convergence. This method not only maintains the collaborative consistency between UAVs and unmanned vehicles in complex environments but also possesses adaptive repair capabilities, allowing for rapid recovery of stable paths under conditions of sudden local environmental changes or communication fluctuations. This improves overall delivery success rate and system reliability. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0016] Figure 1 This is a flowchart of a collaborative delivery route planning method for drones and unmanned vehicles according to the present invention.

[0017] Figure 2 This is a flowchart of the method for determining the stability of P according to the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] For examples, please refer to Figure 1 , 2 As shown in this embodiment, a method for collaborative delivery route planning using drones and unmanned vehicles includes: The time-varying micro-airflow field, dynamic constraint area, and communication accessibility parameters of the delivery area are obtained to construct a cross-domain environmental parameter set E.

[0020] In this embodiment, firstly, the delivery area is spatially gridded and discretized, dividing the target area into several three-dimensional unit grids, and a unique spatial index is established for each grid. Based on this, by using barometric pressure sensors, wind speed and direction sensors, and ground meteorological monitoring terminals deployed on the UAV, combined with historical meteorological data and short-term prediction models, the airflow velocity, direction, and disturbance intensity of each grid at different time slices are obtained, thereby forming micro-airflow time-varying field data with a time dimension. The micro-airflow time-varying field includes not only average wind field information, but also local turbulence intensity parameters caused by building gaps, canyon topography, or thermal differences.

[0021] To obtain dynamic constraint zones, the system integrates traffic management system data, temporary control information, and environmental perception results to identify restricted access areas within the delivery area that change over time. Specifically, the dynamic constraint zones on the unmanned vehicle side include temporary road closures, congested road sections, and construction areas, while the dynamic constraint zones on the drone side include temporary no-fly zones, areas with sudden obstacles, and low-altitude control areas. These constraint zones are then uniformly mapped onto the spatial grid to form accessibility marker parameters that are updated over time.

[0022] For communication accessibility parameters, signal strength, link stability, and data packet loss rate of drones and unmanned vehicles at different locations are collected, and communication attenuation distribution in the delivery area is constructed by combining existing communication base station distribution and obstruction models. The communication accessibility parameters include at least signal strength threshold, link interruption probability, and delay estimate, and are also associated with the spatial grid and time slice.

[0023] The time-varying field parameters of micro-airflow, dynamic constraint zone parameters, and communication accessibility parameters are integrated to quantify the impact of different types of environmental factors on the collaborative path of UAVs and unmanned vehicles, forming a cross-domain environmental parameter set E. The set E is used to characterize the coupling environmental characteristics between the airspace and ground domain within the delivery area and serves as the basic input for subsequent collaborative path planning and mismatch propagation analysis.

[0024] Based on the delivery task and vehicle status, establish a handover and coupling relationship diagram G between drones and unmanned vehicles.

[0025] In this embodiment, firstly, based on the handover requirements and vehicle operating status in the delivery task, a set of candidate handover nodes for drones and unmanned vehicles is extracted. Specifically, the delivery area is divided into discrete spatial grids, and grid cells that simultaneously satisfy the conditions for drone landing and unmanned vehicle docking are selected as candidate handover nodes. The conditions for drone landing are defined as the height of obstacles in the corresponding airspace of the grid being lower than a preset safe height threshold H1, and the horizontal wind speed being lower than a preset threshold V1. The conditions for unmanned vehicle docking are defined as the ground slope of the corresponding grid being lower than a threshold θ1, and the passability being accessible. Subsequently, based on the time window constraints of the delivery task, time matching is performed on the candidate handover nodes. Specifically, the expected arrival time tα of the drone and the expected arrival time tβ of the unmanned vehicle are calculated, and a time difference threshold ΔT1 is set. When |tα−tβ|≤ΔT1, it is determined to be a matchable node, thereby forming an initial pair of handover nodes.

[0026] For the initial handover node pair, a feasible handover constraint relationship between nodes is constructed by combining the vehicle's remaining energy, load capacity, and operational constraints. Specifically, a handover cost function C(i,j) = αEd + βEu + γL is defined, where Ed is the energy consumption of the unmanned vehicle traveling to the node, Eu is the energy consumption of the UAV flying to the node, L is the waiting time loss, and α, β, and γ are weight coefficients, whose values ​​are determined by normalizing historical operational data. When both Ed and Eu are less than their respective remaining energy thresholds and the load meets the constraints, a connection relationship between nodes is established, and the handover cost function value is used as the weight of the connection edge to form a weighted connection edge set.

[0027] Based on the weighted connection edges, a temporal consistency constraint is introduced to align the arrival times of the UAV and the unmanned vehicle. Specifically, using the estimated arrival time of the unmanned vehicle as a benchmark, the UAV flight path is fine-tuned, the adjusted arrival time tα′ is calculated, and a temporal offset function is defined. Set a timing consistency threshold ΔT2. When Δt≤ΔT2, retain the connection edge; otherwise, discard it. The flight path fine-tuning is achieved by adjusting the flight speed or inserting buffer segments within the allowable range.

[0028] Finally, based on the set of connecting edges satisfying the temporal consistency constraint and the corresponding handover nodes, a handover coupling graph with spatiotemporal coupling properties is constructed. Specifically, each handover node is defined as a vertex in the graph, the connecting edges satisfying the constraints are used as an edge set, and the handover cost, temporal offset, and energy consumption parameters are recorded in the edge attributes, thus forming a complete handover coupling graph for subsequent path optimization and collaborative scheduling calculations.

[0029] Based on E and G, a unified spatiotemporal evolution model W is constructed, and a temporal phase synchronization constraint is introduced to ensure that the UAV flight trajectory and the unmanned vehicle driving trajectory satisfy the phase alignment relationship at the handover node.

[0030] In this embodiment, firstly, the spatial grid and time series information in the cross-domain environmental parameter set are mapped to nodes and connecting edges in the intersection coupling relationship graph. Specifically, using the three-dimensional spatial grid in the cross-domain environmental parameter set as the basic unit, the environmental parameters of each grid at different time slices are attached to the corresponding spatial location; simultaneously, each intersection node in the intersection coupling relationship graph is located to the corresponding grid and assigned a time label tk, which is discretely divided according to a fixed time step Δτ; the connecting edges inherit the time interval information of the starting node and the ending node, thereby forming an extended spatiotemporal node set, where each node contains spatial coordinates, a time label, and environmental constraint parameters, realizing a unified expression of space and time.

[0031] Based on the extended spatiotemporal node set, UAV flight trajectory functions and unmanned vehicle (UAV) driving trajectory functions are established. Specifically, the UAV flight trajectory function is defined as a continuous function fu(t)=(xu(t),yu(t),zu(t)), and the UAV driving trajectory function is defined as fv(t)=(xv(t),yv(t)). The time interval is discretized according to the step size Δτ to obtain the time sequence {t1,t2,…,tn}, and the spatial position corresponding to each time point is calculated to form a discrete trajectory sequence. The trajectory generation is achieved by combining shortest path search with environmental constraints, specifically using a weighted cost search method. During the search process, micro-airflow disturbance, dynamic constraints, and communication reachability are accumulated as components of the cost function. Then, a temporal phase function is constructed based on the trajectory sequence. Specifically, for each handover node i, the UAV arrival time tu(i) and the UAV arrival time tv(i) are calculated, and a temporal phase function is defined. Simultaneously, a phase difference threshold ΔΦ is set, which is determined based on historical task statistics. Specifically, the threshold is determined by statistically analyzing the time deviation distribution in successful handover tasks and taking its percentile as the threshold. This timing phase function is used to quantify the synchronization degree between the UAV and the unmanned vehicle at the handover node.

[0032] The trajectory sequences are adjusted and filtered based on the phase difference threshold. Specifically, when the temporal phase function value φ(i) of a certain handover node is greater than the phase difference threshold ΔΦ, the UAV flight trajectory function or the unmanned vehicle driving trajectory function is time-corrected. The time correction is achieved by adjusting the UAV flight speed or the unmanned vehicle driving speed, wherein the speed adjustment range is limited to a preset speed range. After correction, the temporal phase function value of the corresponding node is recalculated until φ(i)≤ΔΦ is satisfied. Trajectory combinations that cannot meet the condition through speed adjustment are eliminated. Finally, all trajectory sequences that satisfy the phase alignment relationship are retained, and a unified spatiotemporal evolution model is constructed based on this model for subsequent collaborative path optimization calculations.

[0033] In W, a mismatch propagation function F is constructed to map the trajectory offset generated by the UAV to the path disturbance response of the unmanned vehicle, thus obtaining the initial cooperative path set P.

[0034] In this embodiment, firstly, the trajectory offset of the UAV is extracted at the junction nodes in the unified spatiotemporal evolution model, and a UAV side offset feature sequence is constructed. Specifically, in the unified spatiotemporal evolution model, for each junction node, the actual spatial position coordinates of the UAV at the corresponding time of that node and the target spatial position coordinates of the corresponding node in the planned path are obtained; wherein, the actual spatial position coordinates are obtained through UAV positioning data, and the planned spatial position coordinates are determined by the path planning results. Subsequently, the difference between the actual position and the planned position is quantified using three-dimensional Euclidean distance as a metric, that is, by squaring the position differences in the three coordinate directions, summing them, and then taking the square root, the trajectory offset at that node is obtained. The trajectory offset reflects the degree of deviation of the UAV due to environmental disturbances or control errors during flight; the larger the value, the more serious the deviation of the actual trajectory from the planned trajectory. Subsequently, the trajectory offsets of adjacent nodes are differentially calculated according to the time series to obtain the offset change rate. , where Δτ is the time step; arrange the offset change rates corresponding to all handover nodes in time order to form the UAV side offset feature sequence.

[0035] Secondly, a mismatch propagation function is constructed based on the offset feature sequence. Specifically, the mismatch propagation function is defined as follows: , where ru(i) is the offset change rate at node i, Δt(i,j) is the time interval between node i and node j, and κ(i,j) is the environmental disturbance coefficient of the path segment. This disturbance coefficient is calculated by normalizing the micro-airflow intensity and communication attenuation value in the cross-domain environmental parameter set. The weight parameters w1, w2, and w3 are determined by least-squares fitting of historical delivery data to minimize the historical trajectory error in the function output. Through the above function, the UAV-side offset characteristics are mapped to the path disturbance response Δv(i,j) on the UAV side.

[0036] Then, the path disturbance response is superimposed onto the original driving trajectory of the autonomous vehicle to correct the disturbance in the vehicle's path. Specifically, for the position (xv(j), yv(j)) of node j in the autonomous vehicle's trajectory, its spatial offset is adjusted according to Δv(i,j) to obtain the corrected position (xv′(j), yv′(j)), where the adjustment direction is along the tangential direction of the original path, and the adjustment magnitude is Δv(i,j); simultaneously, the corresponding time parameters are updated synchronously, i.e. , where v(j) is the speed of the unmanned vehicle at node j; by performing the above correction operation on all nodes, the set of unmanned vehicle paths after disturbance is obtained.

[0037] Finally, the disturbed unmanned vehicle path and the drone flight trajectory are coupled and matched to generate an initial cooperative path set. Specifically, at each handover node, the arrival time difference between the drone and the unmanned vehicle is recalculated and judged according to the phase difference threshold in the temporal phase synchronization constraint; when the phase alignment condition is met, the corresponding drone trajectory and unmanned vehicle path are combined into a set of effective cooperative paths; after traversing and filtering all nodes, all effective path combinations are summarized to form an initial cooperative path set for subsequent stability analysis and optimization calculations.

[0038] The stability of P is determined, the mismatch accumulation section caused by F is identified, and the set of unstable regions Q is generated.

[0039] In this embodiment, firstly, a mismatch transmission matrix is ​​constructed based on the connection relationships between each handover node in the initial cooperative path set. Specifically, the handover nodes in the path are numbered sequentially as i=1,2,…,N. For any adjacent nodes i and j, matrix elements are defined. Where Δv(i,j) is the path disturbance response of the unmanned vehicle between nodes, and δu(i) is the trajectory offset of the UAV at node i; for non-adjacent nodes, their matrix elements are set to 0, thus forming a sparse matrix; then, the mismatch transfer matrix is ​​decomposed into eigenvalues, and the modulus ρ of its largest eigenvalue is calculated using the power iteration method, which is the spectral radius index; the power iteration method specifically involves: initializing the vector x0, and performing iteration. The process continues until convergence, and the spectral radius value is finally calculated using the Rayleigh quotient. The spectral radius index characterizes the overall amplification capability of mismatches in the path. The mismatch amounts corresponding to each intersection node are mapped to the node attributes of the intersection coupling graph to construct the graph signal, and the total graph variation value is calculated. Specifically: taking the nodes in the intersection coupling graph as vertices, the UAV trajectory offset or UAV disturbance amount corresponding to the node is used as the node signal value s(i); for any connecting edge (i,j), the node signal difference is calculated. The total graph variation is obtained by weighting the graph with edge weights w(i,j). The edge weights are taken from the normalized result of the crossover cost function; the total graph variation value is used to characterize the spatial jump degree of mismatch in the graph structure. The spectral radius index and the total graph variation value are weighted and fused to calculate the path-level mismatch gain coefficient. Specifically, the path-level mismatch gain coefficient is defined as follows: λ1 and λ2 are weighting coefficients, which are determined by fitting the historical path stability samples with the minimum error. This fusion method simultaneously reflects the amplification capability of mismatch in path propagation and spatial discontinuity.

[0040] The temporal phase difference data of each handover node in the initial collaborative path set is obtained, and the path phase entropy value is constructed. Specifically, the temporal phase difference φ(i) of each node is calculated, and it is divided into K intervals according to a preset interval. The interval division method is to divide the phase difference range [0, Φmax] evenly or to divide it non-uniformly according to the historical distribution; the number of nodes nk in each interval is counted, and the probability pk=nk / N is calculated; then, the information entropy formula is applied. Calculate path phase entropy value This is used to characterize the discreteness of timing consistency. Subsequently, the path-level mismatch gain coefficient and the path phase entropy value are jointly mapped to construct a two-dimensional stability judgment space. Specifically, a two-dimensional coordinate system is established with the path-level mismatch gain coefficient as the horizontal axis and the path phase entropy value as the vertical axis, and a stability boundary function is defined. , where a, b, and c are parameters obtained by fitting historical stable and unstable samples; when S(G,H)>0, it is determined to be an unstable region, and when S(G,H)≤0, it is determined to be a stable region.

[0041] Finally, each segment in the path is judged segment by segment. When the path-level mismatch gain coefficient and the path phase entropy value of a certain segment fall into the unstable region, the segment is marked as the mismatch accumulation segment. The above judgment is performed on all path segments, and all marked segments are summarized to form an unstable region set for subsequent path reconstruction and optimization.

[0042] In the region corresponding to Q, path recoverability coding is introduced to perform redundant segmentation and reconstruction marking on the path, and W is locally reconstructed based on the coding to obtain the corrected model W′.

[0043] In this embodiment, firstly, a path recoverability coding sequence is constructed for each path segment in the unstable region set. Specifically, each unstable path segment is numbered according to its position in the initial cooperative path set, and a path segment identifier ID(i) is defined; based on the path-level mismatch gain coefficient and the path phase entropy value, a mismatch level identifier L(i) is defined, wherein the path stability is graded by setting two threshold intervals [G1,G2] and [H1,H2]. If both the path-level mismatch gain coefficient and the path phase entropy value fall into the high interval, it is marked as a high mismatch level; at the same time, all feasible alternative paths connected to the start and end nodes of the segment are searched in the unified spatiotemporal evolution model, and a unique index K(i,j) is assigned to each alternative path; finally, the path segment identifier, mismatch level identifier, and alternative path index are combined to form a path recoverability coding sequence C(i)=[ID(i),L(i),K(i,j)]. Secondly, based on the path recoverability coding sequence, the unstable path segments are subjected to redundancy segmentation processing to generate alternative paths. Specifically, for each path segment in the coding sequence, a constrained path search is performed in a unified spatiotemporal evolution model according to its start and end nodes. The path search adopts a weighted cost search method, where the cost function is defined as follows: Where E represents energy consumption, T represents time delay, D represents environmental disturbance intensity, and μ1, μ2, and μ3 are weighting coefficients determined through historical sample normalization. During the search process, an energy threshold Emax and a time threshold Tmax are set, retaining only paths satisfying E≤Emax and T≤Tmax as candidate paths. For each unstable segment, at least one candidate path satisfying the constraints is generated, forming a redundant path set. Then, based on the alternative path index in the path recoverability encoding sequence, the original path segment is reconstructed, marked, and replaced. Specifically: based on the mismatch level identifier, the candidate path with the smallest cost function value is preferentially selected as the replacement path; in the original path structure, the node sequence of the corresponding unstable segment is replaced with the node sequence of the selected candidate path, while updating the connection relationships and time labels between nodes; for high mismatch level segments, multiple candidate paths can be retained simultaneously to form a branch structure to improve path recovery capability; after the replacement is completed, the updated path structure is obtained.

[0044] Finally, based on the updated path structure, the unified spatiotemporal evolution model is locally reconstructed. Specifically, the nodes and connecting edges corresponding to the original unstable segments are deleted from the model, and new nodes and connecting edges corresponding to the replacement paths are inserted. The newly inserted nodes are reassigned spatial location, time label, and environmental parameters, and the handover cost, temporal constraints, and perturbation parameters of the connecting edges are updated. Subsequently, a local consistency check is performed on the reconstructed region, specifically by recalculating the temporal phase difference and comparing it with a preset threshold to ensure that the phase alignment condition is met. After completing the above processing, a corrected unified spatiotemporal evolution model is formed for subsequent path optimization and execution.

[0045] Based on W′, iterative optimization is performed to ensure that the path meets energy consumption constraints while suppressing mismatch propagation and achieving cross-carrier temporal convergence, outputting the final collaborative delivery path scheme.

[0046] In this embodiment, firstly, a joint optimization objective function is constructed within the modified unified spatiotemporal evolution model. Specifically, the joint optimization objective function is defined as follows: taking all nodes in the path as computational units. Where Etotal represents the total energy consumption of the path, obtained by summing the energy consumption of the UAV flight and the energy consumption of the unmanned vehicle driving; Ftotal represents the mismatch propagation intensity, obtained by summing the output values ​​of the mismatch propagation function of each path segment; Φtotal represents the total temporal phase deviation of the path, obtained by summing the temporal phase differences of each junction node. Next, based on the joint optimization objective function, the path is iteratively updated. Specifically: the current path is initialized with the corrected path structure. In each iteration, the connection order and time labels of the nodes in the path are fine-tuned sequentially. The fine-tuning includes two methods: one is to swap adjacent nodes to change the path topology, and the other is to increase or decrease the node time labels to optimize the temporal distribution. After each adjustment, the objective function value J is recalculated, and the adjustment result that reduces J is retained. The iterative process adopts a gradient approximate descent strategy, that is, the optimization direction is determined by comparing the difference ΔJ of the objective function before and after the adjustment, and the adjustment is accepted when ΔJ < 0. During each iteration, the path constraints are determined. Specifically, the total energy consumption Etotal, mismatch propagation intensity Ftotal, and total temporal phase deviation Φtotal of the path are calculated and compared with preset thresholds. The energy threshold Emax is determined based on the maximum available energy of the vehicle, the mismatch threshold Fmax is determined by statistically analyzing the average mismatch level of historical stable paths, and the phase convergence threshold Φmax is determined by setting the maximum allowable time deviation range. When any indicator exceeds the corresponding threshold, the corresponding path segment is located, and a local adjustment operation is performed on the segment, including reselecting connection edges or reassigning time labels, until the constraints are met.

[0047] Finally, a convergence criterion is set to terminate the iteration process. Specifically, it is determined when the objective function changes over K consecutive iterations. When the value is less than the preset convergence threshold ε, the optimization process is considered to have converged. The convergence threshold ε is determined experimentally to be a value less than one percent of the total path cost. After the convergence condition is met, the current path is output as the final collaborative delivery path scheme. This path simultaneously satisfies the energy consumption constraint, the mismatch propagation suppression requirement, and the timing convergence condition of the UAV and the unmanned vehicle at the handover node.

[0048] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for collaborative delivery route planning using drones and unmanned vehicles, characterized in that: include: Acquire the time-varying micro-airflow field, dynamic constraint area, and communication accessibility parameters of the delivery area to construct a cross-domain environmental parameter set E; Based on the delivery task and vehicle status, establish a handover and coupling relationship diagram G between drones and unmanned vehicles; A unified spatiotemporal evolution model W is constructed based on E and G, and a temporal phase synchronization constraint is introduced to ensure that the UAV flight trajectory and the UAV driving trajectory satisfy the phase alignment relationship at the handover node. In W, a mismatch propagation function F is constructed to map the trajectory offset generated by the UAV to the path disturbance response of the unmanned vehicle, thus obtaining the initial cooperative path set P; The stability of P is determined, the mismatch accumulation section caused by F is identified, and the set of unstable regions Q is generated. In the region corresponding to Q, path recoverability coding is introduced to perform redundant segmentation and reconstruction marking on the path, and W is locally reconstructed based on the coding to obtain the corrected model W′. Based on W′, iterative optimization is performed to ensure that the path meets energy consumption constraints while suppressing mismatch propagation and achieving cross-carrier temporal convergence, outputting the final collaborative delivery path scheme.

2. The method for collaborative delivery path planning of unmanned aerial vehicles and unmanned vehicles according to claim 1, characterized in that: The establishment of the handover and coupling relationship diagram G between the UAV and the unmanned vehicle includes the following steps: Based on the handover requirements and vehicle operation status in the delivery task, a set of candidate handover nodes for drones and unmanned vehicles is extracted, and the candidate handover nodes are matched and filtered based on spatial location and time window to form an initial handover node pair; For the initial handover node pair, based on the vehicle's remaining energy, load capacity, and operational constraints, a feasible handover constraint relationship between the nodes is constructed, and weighted connection edges are generated accordingly. Based on the weighted connection edges, a temporal consistency constraint is introduced to perform alignment calculations on the arrival time of the UAV and the arrival time of the unmanned vehicle, and to determine the set of valid edges that meet the synchronous handover conditions. Based on the set of effective edges and the corresponding intersection nodes, an intersection coupling relationship graph with spatiotemporal coupling properties is constructed.

3. The method for collaborative delivery path planning of unmanned aerial vehicles and unmanned vehicles according to claim 1, characterized in that: The construction of a unified spatiotemporal evolution model W based on E and G includes the following steps: The spatial grid and time series information in the cross-domain environmental parameter set are mapped to the nodes and connecting edges in the intersection and coupling relationship graph, an extended spatiotemporal node set containing the time dimension is constructed, and each node is assigned environmental constraint attributes and time markers. Based on the extended spatiotemporal node set, a UAV flight trajectory function and an unmanned vehicle driving trajectory function are established, and the trajectories are sampled by discrete time steps to form corresponding trajectory sequences. A temporal phase function is constructed based on the trajectory sequence. The temporal phase function is used to characterize the difference in arrival stages between the UAV and the unmanned vehicle at the same handover node, and a phase difference threshold is defined to constrain the degree of synchronization between the two. The trajectory sequences are adjusted and filtered according to the phase difference threshold, and the trajectory combinations that satisfy the phase alignment relationship are retained, thereby forming a unified spatiotemporal evolution model.

4. The method for collaborative delivery path planning of unmanned aerial vehicles and unmanned vehicles according to claim 1, characterized in that: The process of mapping the trajectory offset generated by the UAV to the path disturbance response of the vehicle to obtain the initial cooperative path set P includes the following steps: The trajectory offset of the UAV is extracted at the junction nodes in the unified spatiotemporal evolution model, and the continuous offset change rate is calculated based on the time series to form the UAV side offset feature sequence. A mismatch propagation function is constructed based on the offset feature sequence. The mismatch propagation function takes the offset change rate and the time interval between nodes as input and calculates the path disturbance response of the UAV through weighted mapping. The path disturbance response is superimposed on the original driving trajectory of the UAV to correct the disturbance of the UAV path and obtain the disturbed path set. The disturbed UAV path is coupled and matched with the UAV flight trajectory to select the path combination that meets the temporal phase synchronization constraint, thereby generating the initial cooperative path set.

5. The method for collaborative delivery route planning of unmanned aerial vehicles and unmanned vehicles according to claim 1, characterized in that: The stability determination of P includes the following steps: For each handover node in the initial cooperative path set, calculate the path-level mismatch gain coefficient; Based on the temporal phase difference data of each handover node, a probability distribution is constructed according to the preset interval division rules, and the path phase entropy value is obtained through the information entropy calculation method. The path-level mismatch gain coefficient and the path phase entropy value are jointly mapped to construct a two-dimensional stability determination space, and each path segment is classified and labeled according to a preset stability boundary function. Path segments that are deemed not to meet stability boundary conditions are identified as mismatch accumulation segments and aggregated to form a set of unstable regions.

6. The method for collaborative delivery path planning of unmanned aerial vehicles and unmanned vehicles according to claim 5, characterized in that: The calculation of the path-level mismatch gain coefficient includes the following steps: Based on the connection relationship between each handover node in the initial cooperative path set, a mismatch transfer matrix is ​​constructed. The matrix elements of the mismatch transfer matrix are determined by the ratio of the unmanned vehicle path disturbance response between adjacent nodes to the UAV trajectory offset. The mismatch transfer matrix is ​​decomposed into eigenvalues, and the magnitude of its largest eigenvalue is calculated as the spectral radius index. The spectral radius index is then used as a characterization parameter of the overall mismatch amplification capability of the path. Map the mismatch corresponding to each junction node to the node attributes of the junction coupling relationship graph to construct the graph signal, and calculate the total graph variation value based on the connection edges between nodes; The path-level mismatch gain coefficient is obtained by weighted fusion calculation of the spectral radius index and the total variation value of the graph.

7. The method for collaborative delivery path planning of unmanned aerial vehicles and unmanned vehicles according to claim 5, characterized in that: The calculation of the path phase entropy value includes the following steps: The timing phase difference data of each handover node in the initial collaborative path set is obtained, and the timing phase difference is segmented according to the preset interval division rule. The preset interval is divided into equal or non-equal intervals according to the phase difference value range to form multiple phase intervals. The number of nodes in each phase interval is counted, and the probability distribution value of each interval is calculated by the proportion of the number of nodes to the total number of nodes, thereby constructing the phase probability distribution sequence of the path; Based on the phase probability distribution sequence, the probability values ​​of each interval are weighted and summed according to the information entropy calculation formula to obtain the path phase entropy value.

8. The method for collaborative delivery route planning of unmanned aerial vehicles and unmanned vehicles according to claim 1, characterized in that: The step of introducing path recoverability coding in the Q-corresponding region and performing redundancy segmentation and reconstruction marking on the path includes the following steps: For each path segment in the set of unstable regions, a path recoverability coding sequence is constructed. The path recoverability coding sequence consists of a path segment identifier, a mismatch level identifier, and an alternative path index to characterize the reconstruction priority and replacement relationship of each segment. Based on the path recoverability coding sequence, the unstable path segment is subjected to redundant segmentation processing. At least one alternative path is generated in each segment, and a set of redundant paths that meet the energy consumption and timing requirements is selected through constraints. Based on the substitution relationship in the path recoverability coding sequence, the original path segment is reconstructed and marked, and the selected redundant path is embedded into the original path structure to form an updated path connection relationship. Based on the updated path connection relationship, the corresponding nodes and connecting edges in the unified spatiotemporal evolution model are locally replaced and reconstructed to obtain the corrected unified spatiotemporal evolution model.

9. The method for collaborative delivery path planning of unmanned aerial vehicles and unmanned vehicles according to claim 8, characterized in that: The iterative optimization based on W′ includes the following steps: In the modified unified spatiotemporal evolution model, a joint optimization objective function is constructed. The joint optimization objective function consists of an energy consumption term, a mismatch propagation suppression term, and a time phase convergence term. Each term is linearly combined through weighted coefficients to form a comprehensive evaluation index. Based on the joint optimization objective function, the path is iteratively updated and calculated. A combination of gradient descent and path rearrangement is used to gradually adjust the node connection order and time label. In each iteration, the energy consumption, mismatch propagation intensity and phase difference change of the path are calculated and compared with the preset energy consumption threshold, mismatch threshold and phase convergence threshold. When any constraint is not satisfied, the corresponding path segment is locally corrected. When the results of multiple consecutive iterations meet the convergence criteria, the iteration process stops, and the final collaborative delivery path scheme that satisfies energy consumption constraints and achieves cross-carrier temporal convergence is output.

Images