Logistics distribution path planning method and system based on drone swarm

Abstract

The present invention provides a logistics distribution path planning method and system based on a drone cluster, which relates to the field of drone distribution technology. First, logistics task data and environmental monitoring data of a target area are acquired, and then feature extraction is performed on the logistics task data and the environmental monitoring data to obtain a path planning feature set and an environmental constraint feature set. Path optimization processing is performed based on the path planning feature set and the environmental constraint feature set to generate a global path plan and a dynamic adjustment plan. Subsequently, the generated global path plan and dynamic adjustment plan are verified to obtain a verification result set. When the verification result meets the execution condition, the global path plan and the dynamic adjustment plan are sent to the drone cluster for execution, thereby comprehensively considering logistics tasks and environmental factors, making the planning scientific and reasonable, and having a dynamic adjustment and verification mechanism, thereby improving the efficiency and reliability of drone cluster logistics distribution.

Description

Technical Field

[0001] The present invention relates to the field of drone delivery technology, and in particular to a logistics delivery path planning method and system based on drone clusters. Background Art

[0002] In today's rapidly developing logistics industry, drone swarm logistics delivery, with its high efficiency and flexibility, is gradually becoming an important means to solve the "last mile" delivery problem. Traditional drone logistics delivery path planning methods have many shortcomings.

[0003] On the one hand, some existing methods only consider basic logistics task information, such as the origin and destination of a delivery, while ignoring the priority of delivery tasks. This can result in important and urgent tasks not being prioritized during actual delivery, reducing the quality and efficiency of logistics services and failing to meet the needs of some customers with high timeliness requirements.

[0004] On the other hand, existing technologies tend to take a simplistic and one-sided approach to environmental factors when planning routes. Some methods focus solely on terrain data, ignoring the impact of weather conditions on drone flight. For example, severe weather conditions like strong winds and heavy rain can hinder drone flight and even lead to safety accidents. Conversely, considering only weather data without considering terrain can also lead to inappropriate route planning. For example, in complex terrain such as mountainous areas, routes unsuitable for drone flight may be planned.

[0005] Furthermore, existing path planning systems mostly generate fixed, global routing solutions and lack dynamic adjustment mechanisms. This makes it difficult to respond and adjust to sudden environmental changes or unexpected logistics tasks, significantly reducing the efficiency and reliability of drone swarm delivery. Summary of the Invention

[0006] In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides a logistics distribution path planning method based on a drone cluster, the method comprising:

[0007] Obtaining logistics task data and environmental monitoring data for the target area, wherein the logistics task data includes a set of delivery starting points, delivery destinations, and delivery task priorities, and the environmental monitoring data includes meteorological data and terrain data;

[0008] Performing feature extraction on the logistics task data and the environmental monitoring data to generate a path planning feature set and an environmental constraint feature set;

[0009] Performing path optimization processing based on the path planning feature set and the environmental constraint feature set to generate a global path plan and a dynamic adjustment plan for the UAV cluster;

[0010] Performing verification processing on the global path solution and the dynamic adjustment solution to generate a verification result set;

[0011] When the verification result set meets the execution conditions, the global path plan and the dynamic adjustment plan are sent to the drone cluster for execution.

[0012] On the other hand, an embodiment of the present invention also provides a logistics distribution path planning system based on a drone cluster, including a processor and a machine-readable storage medium, wherein the machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0013] Based on the above aspects, the embodiment of the present invention obtains logistics task data of the target area including a delivery starting point, a delivery destination set, and a delivery task priority, as well as environmental monitoring data such as meteorological data and terrain data, generates a path planning feature set and an environmental constraint feature set through feature extraction, performs path optimization processing to obtain a global path plan and a dynamic adjustment plan, and then performs verification processing to obtain a verification result set. When the execution conditions are met, it is sent to the drone cluster for execution. This comprehensively integrates logistics tasks and environmental factors from a holistic perspective, avoids the problem of unreasonable path planning caused by considering only a single factor, and greatly improves the scientificity and rationality of path planning; by taking the logistics task priority into consideration, it can prioritize the efficient delivery of important tasks and improve the service quality of logistics delivery; the generated dynamic adjustment plan enables the drone cluster to adjust the path in a timely manner when encountering complex and changing environmental conditions, enhances the ability to cope with uncertainty, and improves the reliability and success rate of delivery; the verification processing link ensures that the generated path plan meets the actual execution requirements, reduces execution risks, reduces resource waste caused by path planning errors, and overall improves the efficiency and safety of drone cluster logistics delivery. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a schematic diagram of the execution flow of the logistics distribution path planning method based on drone clusters provided by an embodiment of the present invention.

[0015] Figure 2 Schematic diagram of exemplary hardware and software components of a logistics distribution path planning system based on drone swarms provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0016] The present invention will be described in detail below with reference to the accompanying drawings. Figure 1This is a flow chart of a logistics distribution path planning method based on drone clusters provided by an embodiment of the present invention. The logistics distribution path planning method based on drone clusters is introduced in detail below.

[0017] Step S110: Obtaining logistics task data and environmental monitoring data of the target area, wherein the logistics task data includes a delivery starting point, a delivery destination set, and a delivery task priority, and the environmental monitoring data includes meteorological data and terrain data.

[0018] In the scenario set in this embodiment, there is a specific target area for logistics distribution services. For logistics task data, the delivery starting point is marked as A, and its coordinates are represented by (x_A, y_A). These two coordinate values ​​determine the position of the starting point in plane space. The set of delivery destinations is represented as {B1, B2, ..., Bn}, where each delivery destination Bi has corresponding coordinates (x_Bi, y_Bi) to clearly define the location of each destination. The delivery task priority consists of a set {Q1, Q2, ..., Qn}, where Qi corresponds to the delivery destination Bi and is used to reflect the priority of each delivery task. For example, the priority can be determined based on factors such as the delivery time required by the customer.

[0019] Regarding environmental monitoring data, meteorological data includes elements such as wind speed, rainfall, and temperature. Wind speed data is organized chronologically into a sequence V = {v1, v2, …, vt}, where vi represents the wind speed at the i-th time point. Rainfall data is similarly organized chronologically into a sequence R = {r1, r2, …, rt}, where ri represents the rainfall at the i-th time point. Temperature data is organized into a sequence T = {t1, t2, …, tt}, where ti represents the temperature at the i-th time point. Topographic data is presented in the form of a digital elevation model, assuming that the model consists of multiple grid cells, with each grid cell g having a corresponding elevation value h_g.

[0020] Step S120: extracting features from the logistics task data and the environmental monitoring data to generate a path planning feature set and an environmental constraint feature set.

[0021] Next, feature extraction operations are performed on the acquired logistics task data and environmental monitoring data.

[0022] Step S121: segment the delivery starting point and the delivery destination set according to the delivery task priority, and generate multiple sub-path segments and corresponding time constraint parameters as the path planning feature set, each sub-path segment includes starting coordinates and ending coordinates.

[0023] In this scenario, the delivery starting point A and the delivery destination set {B1, B2, ..., Bn} are segmented according to the delivery task priority Qi. For example, if Q1 is the highest priority, then the path from the delivery starting point A to B1 is a sub-path segment. Let this sub-path segment be SP1, and its starting coordinates are the coordinates of A (x_A, y_A), and the ending coordinates are the coordinates of B1 (x_B1, y_B1). The determination of the time constraint parameter is based on multiple factors. Assuming that the delivery distance and the nature of the goods are taken into account, let the distance from A to B1 be d_A_B1. According to the preset speed rule v_rule and the maximum transportation time limit t_limit allowed for the goods, the time constraint parameter t_c1 is calculated. The calculation logic is:

[0024] First, calculate the distance according to the distance formula, which is as follows:

[0025] d_A_B1=[(x_B1-x_A)^2+(y_B1-y_A)^2]^(1 / 2)

[0026] Then, based on the speed rule v_rule (if v_rule is a constant), the estimated travel time t_pre = d_A_B1 / v_rule. Combined with the maximum allowable transport time limit t_limit, t_c1 = min(t_pre, t_limit). Similarly, for each delivery destination Bi, corresponding sub-path segments SPi are generated, with starting coordinates (x_A, y_A), ending coordinates (x_Bi, y_Bi), and a time constraint parameter t_ci. These sub-path segments and their time constraint parameters together constitute the path planning feature set.

[0027] Step S122: Perform regional impact analysis on the meteorological data to generate a set of meteorological constraint parameters.

[0028] In this scenario, regional impact analysis is performed based on meteorological data.

[0029] Step S1221: Divide the meteorological data into multiple meteorological periods according to the time dimension, each meteorological period includes a wind speed sequence, a rainfall sequence and a temperature sequence.

[0030] Assume that the entire time span is divided into k meteorological periods, denoted as T1, T2, …, Tk. For the jth meteorological period Tj, the wind speed sequence is represented as V_j = {v_j1, v_j2, …, v_jm}, where v_ji is the wind speed at the i-th time point within the period; the rainfall sequence is R_j = {r_j1, r_j2, …, r_jm}, where r_ji is the rainfall at the i-th time point within the period; and the temperature sequence is T_j = {t_j1, t_j2, …, t_jm}, where t_ji is the temperature at the i-th time point within the period. Here, m represents the number of time points within each meteorological period, which can be determined based on the actual data collection frequency.

[0031] Step S1222: Perform peak extraction processing on the wind speed sequence of each meteorological period to generate the maximum wind speed value and the average wind speed value, and calculate the wind direction intensity parameter of each period according to the UAV aerodynamic model. The wind direction intensity parameter is defined as the cosine value of the absolute value of the angle between the UAV flight direction and the wind direction.

[0032] For the wind speed sequence V_j={v_j1, v_j2,…, v_jm} of the jth meteorological period, find the maximum value among them by comparison, which is the maximum wind speed value of the period, recorded as v_max_j. The average wind speed value is calculated by adding all the wind speed values ​​in the period and dividing by the number of time points m, that is, v_avg_j=(v_j1+v_j2+…+v_jm) / m. Regarding the calculation of the wind direction intensity parameter, assuming that the flight direction vector of the drone in the period is known to be F_j=(x_Fj, y_Fj), and the wind direction vector is W_j=(x_Wj, y_Wj), according to the vector angle formula, the angle θ is obtained, then the wind direction intensity parameter w_j=|cosθ|, the vector angle formula is as follows:

[0033] cosθ=(x_Fj*x_Wj+y_Fj*y_Wj) / ((x_Fj^2+y_Fj^2)^(1 / 2)*(x_Wj^2+y_Wj^2)^(1 / 2)).

[0034] Step S1223: performing cumulative calculation processing on the rainfall sequence in each meteorological period to generate rainfall range parameters and rainfall duration parameters.

[0035] Taking the rainfall sequence R_j = {r_j1, r_j2, …, r_jm} for the jth meteorological period as an example, the cumulative rainfall is calculated by summing all the rainfall amounts within that period, i.e., r_cum_j = r_j1 + r_j2 + … + r_jm. This r_cum_j serves as a measure of the rainfall range parameter. To determine the rainfall duration parameter, count the number of time points in that period where rainfall is greater than zero, assuming it is n_j. The rainfall duration parameter t_r_j = n_j * Δt, where Δt is the time interval between each time point.

[0036] Step S1224: performing difference calculation processing on the temperature sequence of each meteorological period to generate temperature change parameters and extreme temperature parameters.

[0037] For the temperature sequence T_j={t_j1, t_j2, …, t_jm} of the jth meteorological period, calculate the temperature difference between adjacent time points to obtain the difference sequence ΔT_j={t_j2-t_j1, t_j3-t_j2, …, t_jm-t_j(m-1)}. Perform statistical processing on these differences, such as calculating the average, to obtain the temperature change parameter t_var_j:

[0038] t_var_j=(t_j2-t_j1+t_j3-t_j2+…+t_jm-t_j(m-1)) / (m-1).

[0039] At the same time, find the maximum temperature t_max_j and the minimum temperature t_min_j during this period as the extreme temperature parameters.

[0040] Step S1225: Integrate the wind direction intensity parameter, the rainfall range parameter, the rainfall duration parameter, the temperature change parameter and the extreme temperature parameter into the meteorological constraint parameter set in chronological order.

[0041] In chronological order, the wind direction intensity parameter w_1, rainfall range parameter r_cum_1, rainfall duration parameter t_r_1, temperature variation parameter t_var_1, maximum temperature t_max_1, and minimum temperature t_min_1 of the first meteorological period are combined to form the meteorological constraint parameter group corresponding to the first meteorological period. Similarly, the corresponding parameters w_2, r_cum_2, t_r_2, t_var_2, t_max_2, and t_min_2 of the second meteorological period are combined to form the meteorological constraint parameter group corresponding to the second meteorological period. Similarly, by arranging the meteorological constraint parameter groups of all k meteorological periods in chronological order, a meteorological constraint parameter set is formed.

[0042] Step S123: performing terrain connectivity analysis on the terrain data to generate a set of terrain constraint parameters.

[0043] In this scenario, terrain connectivity analysis is performed on terrain data.

[0044] Step S1231: converting the terrain data into an elevation grid map, and performing region division processing on the elevation grid map to generate multiple connected regions.

[0045] Assume that the original terrain data is converted into an elevation grid map using a specific algorithm. This grid map consists of multiple grid cells. Using a region partitioning algorithm, such as a conventional algorithm based on graph theory, the elevation grid map is processed to partition the grid map into l connected regions, denoted by C1, C2, …, Cl. Each connected region consists of a set of grid cells. For example, connected region C1 contains the grid cells {g11, g12, …, g1s}.

[0046] Step S1232: performing statistical calculations on the elevation grid map of each connected area to generate average elevation parameters, elevation fluctuation parameters, and maximum elevation difference parameters.

[0047] Taking the connected area C1 as an example, to calculate the average elevation parameter, assume that the elevation values ​​of all grid cells in the connected area C1 are h_g11, h_g12, …, h_g1s, then the average elevation parameter h_avg_1=(h_g11+h_g12+…+h_g1s) / s. To calculate the elevation fluctuation parameter, first calculate the square of the difference between the elevation value of each grid cell and the average elevation value h_avg_1, obtaining {(h_g11-h_avg_1)^2, (h_g12-h_avg_1)^2, …, (h_g1s-h_avg_1)^2}. Then, average these squared values ​​and take the square root, which gives the elevation fluctuation parameter h_var_1 = [((h_g11-h_avg_1)^2 + (h_g12-h_avg_1)^2 + … + (h_g1s-h_avg_1)^2) / s]^(1 / 2). To calculate the maximum elevation difference parameter, find the maximum elevation value h_max_1 and the minimum elevation value h_min_1 within the area. The maximum elevation difference parameter h_diff_1 = h_max_1-h_min_1. Similarly, for other connected areas Ci, the average elevation parameter h_avg_i, elevation fluctuation parameter h_var_i, and maximum elevation difference parameter h_diff_i can be obtained.

[0048] Step S1233: performing mutation point identification processing on the elevation grid map according to a preset elevation mutation threshold to generate a coordinate set of terrain obstacle areas.

[0049] Assuming the preset elevation mutation threshold is th, for each grid cell g in the elevation grid map, check the elevation difference between it and its adjacent grid cells. Let the elevation value of grid cell g be h_g, and the elevation value of its adjacent grid cell g' be h_g'. If |h_g - h_g'| > th, grid cell g is considered a mutation point. Record the coordinates of all identified mutation points to form a coordinate set for the terrain obstacle area. For example, if mutation points p1 (with coordinates (x_p1, y_p1)) and p2 (with coordinates (x_p2, y_p2)) are identified, the coordinates of these points constitute part of the coordinate set for the terrain obstacle area.

[0050] Step S1234: Integrate the average elevation parameter, the elevation fluctuation parameter, the maximum elevation difference parameter, and the terrain obstacle area coordinate set into the terrain constraint parameter set.

[0051] The average elevation parameter h_avg_1, elevation fluctuation parameter h_var_1, and maximum elevation difference parameter h_diff_1 of connected region C1 are combined with the coordinate points in the corresponding terrain obstacle region coordinate set to form the terrain constraint parameter set for connected region C1. Similarly, the corresponding parameters of other connected regions, such as C2, are combined and arranged in the order of connected regions to form the terrain constraint parameter set.

[0052] Step S124: combining the meteorological constraint parameter set and the terrain constraint parameter set to generate the environmental constraint feature set.

[0053] In this scenario, the previously generated meteorological constraint parameter set and terrain constraint parameter set are concatenated. For example, the parameter group for the first meteorological period in the meteorological constraint parameter set is concatenated with the terrain constraint parameter group for connected area C1 in the terrain constraint parameter set to form the first combined parameter group. Next, the parameter group for the second meteorological period in the meteorological constraint parameter set is concatenated with the terrain constraint parameter group for connected area C2 in the terrain constraint parameter set to form the second combined parameter group. Similarly, after concatenating the parameter groups for all meteorological periods and corresponding connected areas, an environmental constraint feature set is generated.

[0054] Step S130: Execute path optimization processing based on the path planning feature set and the environmental constraint feature set to generate a global path plan and a dynamic adjustment plan for the UAV cluster.

[0055] In the scenario of this embodiment, path optimization processing is performed based on the generated path planning feature set and environmental constraint feature set.

[0056] Step S131: Construct static constraint conditions based on the sub-path segments and time constraint parameters in the path planning feature set, and dynamically correct the preset cruising speed of the UAV in combination with the wind speed parameter in the real-time meteorological data.

[0057] Taking a certain sub-path segment SPi in the path planning feature set as an example, its starting coordinates are (x_A, y_A), the ending coordinates are (x_Bi, y_Bi), and the time constraint parameter is t_ci. Static constraint conditions are constructed based on this information, that is, it is required that the UAV fly from (x_A, y_A) to (x_Bi, y_Bi) within t_ci time. At the same time, in combination with the wind speed parameter in the real-time meteorological data, let the real-time wind speed be v_real. Assuming the relationship between the performance of the UAV and the influence of meteorological conditions on the speed, let the speed correction function be f(v_real), then the dynamically corrected preset cruising speed v_mod = f(v_real). For example, if f(v_real) = v_pre - k * v_real (where v_pre is the initial preset cruising speed and k is a coefficient determined according to the performance of the UAV), then calculate v_mod according to this formula.

[0058] Step S132: Generate dynamic constraint conditions based on the corrected preset cruising speed, the meteorological constraint parameter set, and the terrain constraint parameter set.

[0059] The corrected cruising speed is v_mod. Combining the meteorological constraint parameter set, such as the wind direction intensity parameter w_j, assuming that when w_j is less than a certain safety value w_safe, in order to ensure flight safety, there are certain restrictions on the flight direction and speed of the UAV. Let the restriction function be g(w_j, v_mod), and determine the flight direction adjustment amount and speed adjustment amount of the UAV according to this function. Then, in combination with the terrain constraint parameter set, such as the terrain obstacle area coordinate set, let the terrain obstacle area be O = {o1, o2,..., on}, where oi is the point coordinate within the obstacle area. When the flight path of the UAV approaches these obstacle areas, according to the distance d from the obstacle area, let the distance calculation function be d(x, y, oi) ((x, y) is the current position coordinate of the UAV), and the preset safety distance d_safe. When d(x, y, oi) < d_safe, adjust the flight path of the UAV. Combining these factors, a series of dynamic constraint conditions are generated.

[0060] Step S133: Call the path optimization algorithm to perform iterative solution processing on the static constraint conditions and the dynamic constraint conditions to generate the global path plan.

[0061] In this scenario, call the path optimization algorithm, such as the genetic algorithm, to solve.

[0062] Step S1331: Initialize a path set including multiple candidate paths, each candidate path including a node sequence and a cost parameter.

[0063] Assume that N candidate paths are initially generated, denoted as P1, P2, …, PN. Taking candidate path P1 as an example, its node sequence can be represented as {n1, n2, …, nm}, where ni is a node in the path, and each node has corresponding coordinates (x_ni, y_ni). The cost parameter is calculated based on multiple factors. Let the weight of the path length factor be w_l, the weight of the time factor be w_t, and the weights of other influencing factors be w_o. To calculate the path length, assume that the distance between adjacent nodes ni and ni+1 is:

[0064] d(ni,ni+1)=[(x_n(i+1)-x_ni)^2+(y_n(i+1)-y_ni)^2]^(1 / 2);

[0065] The total path length L = ∑d(ni, ni+1) (i ranges from 1 to m-1).

[0066] Assume that the time required to traverse the path based on the current speed v_mod is t = L / v_mod. Other influencing factors are set to o, such as the combined impact of weather and terrain conditions. The cost parameter C1 is then calculated as w_l*L+w_t*t+w_o*o. Similarly, for other candidate paths Pi, their node sequences and cost parameters Ci can be obtained.

[0067] Step S1332: Calculate the optimization score of each candidate path according to the static constraint condition, screen the candidate paths according to the dynamic constraint condition, perform genetic operation on the screened candidate paths, and generate an updated path set.

[0068] For example, step S1332-1: obtain the total path length parameter and the total path time parameter corresponding to the node sequence of the candidate path, wherein the total path length parameter is generated by accumulating the Euclidean distances between adjacent nodes, and the total path time parameter is generated by dividing the total path length parameter by the preset cruising speed of the drone cluster.

[0069] In the scenario of this embodiment, for a candidate path P, its node sequence is recorded as {n1, n2, ..., nm}, and each node n i With coordinates (x_n i , y_n). Calculate the total path length parameter L, for the corresponding i Neighbor node n i and n i+1 , the Euclidean distance d(n i , n i+1 ) According to the Euclidean distance calculation formula, we can get: d(ni , n i+1 )=[(x_n( i+1 )-x_n i )^2+(y_n( i+1 )-y_n i )^2]^(1 / 2). Add up the Euclidean distances between all adjacent nodes, that is, the total path length parameter L=∑d(n i , n i+1 ) (i ranges from 1 to m-1).

[0070] Assuming that the preset cruising speed of the UAV cluster is V_pres, the total path time parameter T is obtained by dividing the total path length parameter L by the preset cruising speed V_pres, that is, T=L / V_pres.

[0071] Step S1332-2: Calculate the time deviation parameter according to the static constraint condition, specifically: match the node sequence of the candidate path with the starting coordinates and ending coordinates of each sub-path segment, extract the actual flight time of the corresponding sub-path segment in the candidate path, and calculate the absolute value of the difference between the actual flight time and the time constraint parameter of the sub-path segment.

[0072] Suppose the set of sub-path segments in the path planning feature set is SP={SP1, SP2,…, SPq}, each sub-path segment SPj has a starting coordinate (x_start_j, y_start_j) and an ending coordinate (x_end_j, y_end_j) as well as a time constraint parameter T_const_j.

[0073] For the node sequence of candidate path P, start matching subpath segments from the first node. Assuming that the starting coordinates of subpath segment SPj match at node nk, and the ending coordinates of subpath segment SPj match at node nl (l>k), then the actual flight time T_actual_j of the corresponding subpath segment SPj in the candidate path is calculated based on the total path time parameter calculated previously. The length L_j of the path from nk to nl is also calculated by accumulating the Euclidean distances between adjacent nodes, that is, L_j = ∑d(ni,ni+1) (where i ranges from k to l-1). The actual flight time T_actual_j = L_j / V_pres.

[0074] The time deviation parameter TD_j is the absolute difference between the actual flight time T_actual_j and the time constraint parameter T_const_j of subpath segment SPj, that is, TD_j = |T_actual_j - T_const_j|. This allows us to calculate the corresponding time deviation parameter for each subpath segment, resulting in a set of time deviation parameters TD = {TD1, TD2, …, TDq}.

[0075] Step S1332-3: Calculate the path wind direction influence parameter based on the set of meteorological constraint parameters in the dynamic constraint condition. The path wind direction influence parameter is generated by traversing the node sequence of the candidate path and counting the cumulative sum of the cosine values ​​of the angle between the UAV flight direction and the wind direction during the meteorological period of each node.

[0076] In the meteorological constraint parameter set, for each meteorological period Tj, there is a wind direction intensity parameter W_j. Assume that the flight direction vector of the UAV at each node ni is known to be F_i=(x_Fi, y_Fi), and the wind direction vector for each meteorological period is W_j=(x_Wj, y_Wj).

[0077] Traverse the node sequence of the candidate path P and, for each node ni, determine the meteorological period Tj it is in. Calculate the cosine of the angle cosθ_i between the UAV's flight direction and the wind direction at that node using the vector angle formula cosθ=(x_Fi*x_Wj+y_Fi*y_Wj) / ((x_Fi^2+y_Fi^2)^(1 / 2)*(x_Wj^2+y_Wj^2)^(1 / 2)).

[0078] The path wind direction influence parameter W_sum is obtained by summing the cosine values ​​of the angles at all nodes, that is, W_sum = ∑cosθ_i (i ranges from 1 to m). This path wind direction influence parameter reflects the comprehensive impact of wind direction on the entire candidate path.

[0079] Step S1332-4: Calculate the path obstacle contact parameter according to the terrain constraint parameter set in the dynamic constraint condition, where the path obstacle contact parameter is the number of nodes in the node sequence of the candidate path that fall into the terrain obstacle area coordinate set.

[0080] The terrain constraint parameter set includes a terrain obstacle area coordinate set O={o1, o2, ..., on}, and each coordinate oi=(x_oi, y_oi).

[0081] Traverse the node sequence {n1, n2, …, nm} of the candidate path P and, for each node ni = (x_ni, y_ni), determine whether it falls within the terrain obstacle area coordinate set O. This can be done by calculating the distance d(ni, oi) = [(x_ni - x_oi)^2 + (y_ni - y_oi)^2]^(1 / 2) between node ni and each coordinate oi within the terrain obstacle area and comparing it with a preset distance threshold D_threshold representing the extent of the obstacle area. If d(ni, oi) is less than or equal to D_threshold, node ni is considered to fall within the terrain obstacle area.

[0082] The number of nodes falling into the terrain obstacle area is counted, which is the path obstacle contact parameter No_obs.

[0083] Step S1332-5: performing standardized dimensional conversion on the path total length parameter, the time deviation parameter, the path wind direction influence parameter, and the path obstacle contact parameter to generate a standardized conversion parameter set.

[0084] Since the path total length parameter L, time deviation parameter TD = {TD1, TD2, …, TDq}, path wind direction influence parameter W_sum and path obstacle contact parameter N_obs have different dimensions, standardized dimension conversion is required.

[0085] For the total path length parameter L, it is assumed that there is a maximum path length value L_max (this maximum path length value can be obtained by comparing the total path lengths of all candidate paths), and the standardized total path length parameter L_norm=L / L_max.

[0086] For the time deviation parameter TD_j, it is assumed that there is a maximum time deviation value TD_max (also obtained by comparing the time deviation parameters corresponding to all sub-path segments). The standardized time deviation parameter TD_norm_j = TD_j / TD_max, thus obtaining a set of standardized time deviation parameters TD_norm = {TD_norm1, TD_norm2, …, TD_normq}.

[0087] For the path wind direction influence parameter W_sum, it is assumed that there is a maximum wind direction influence value W_max (obtained by comparing the path wind direction influence parameters of all candidate paths), and the standardized path wind direction influence parameter W_norm=W_sum / W_max.

[0088] For the path obstacle contact parameter N_obs, it is assumed that there is a maximum obstacle contact node number N_max (obtained by comparing the path obstacle contact parameters of all candidate paths), and the standardized path obstacle contact parameter N_norm=N_obs / N_max.

[0089] These standardized parameters are combined together to form a standardized transformation parameter set S = {L_norm, TD_norm, W_norm, N_norm}.

[0090] Step S1332-6: performing weighted summation on each standardized conversion parameter in the standardized conversion parameter set and a predefined weight coefficient to generate an optimization score for each candidate path.

[0091] Assume that the predefined weight coefficients are: path total length weight W_L, time deviation weight W_TD, path wind direction influence weight W_W, and path obstacle contact weight W_N.

[0092] The optimization score Score of the candidate path P is calculated by the following weighted summation formula: Score=W_L*L_norm+∑(W_TD_j*TD_norm_j) (j ranges from 1 to q)+W_W*W_norm+W_N*N_norm.

[0093] The weight coefficient setting here needs to be adjusted according to the actual situation to balance the impact of different factors on path optimization. For example, if time is more important, the time deviation weight W_TD can be set relatively large.

[0094] Step S1332-7: Sort the candidate path set according to the optimization score, retain the candidate paths with optimization scores lower than the preset threshold as the filtered candidate path set, and delete the candidate paths with optimization scores higher than the preset threshold.

[0095] Suppose the candidate path set is P_set={P1, P2,…, PN}, and each candidate path Pi has a corresponding optimization score Score_i.

[0096] A threshold is set, Threshold, and all candidate paths are sorted from smallest to largest by optimization score. After traversing the sorted candidate path set, for candidate paths Pi with an optimization score less than or equal to Threshold, they are retained in the filtered candidate path set P_selected. For candidate paths Pi with an optimization score greater than Threshold, they are removed from the set.

[0097] After this step of screening, a set of relatively better candidate paths P_selected is obtained. These candidate paths are more likely to meet the requirements of path optimization after comprehensively considering factors such as path length, time deviation, wind direction influence and terrain obstacles.

[0098] Step S1332-8: Perform genetic operation processing on the filtered candidate paths to generate an updated path set.

[0099] For example, step S1332-81: randomly select two candidate paths from the filtered candidate path set as parent paths, and select the intersection point according to the following priority:

[0100] If there is a node whose coordinates exist in both parent paths and are not at the starting point or the ending point, then the node is regarded as the intersection point;

[0101] If there is no shared node, randomly select a non-starting or ending node in each of the two parent paths as the crossover point.

[0102] Let the set of candidate paths after screening be P_selected = {P_selected1, P_selected2,..., P_selecteds}. Randomly select two candidate paths from this set. Suppose the selected ones are P_selecteda and P_selectedb.

[0103] First, check the node sequences of P_selecteda and P_selectedb to see if there are node coordinates that exist in both paths and are not at the starting or ending points. Suppose the node sequence of P_selecteda is {n_a1, n_a2,..., n_am}, and the node sequence of P_selectedb is {n_b1, n_b2,..., n_bn}. Traverse the two node sequences. If it is found that there is a node n_ai = n_bj (i≠1 and i≠m, j≠1 and j≠n), then take this node n_ai (or n_bj) as the crossover point.

[0104] If there is no such shared node, randomly select a non-starting or ending node n_ar (1 < r < m) in P_selecteda and a non-starting or ending node n_bs (1 < s < n) in P_selectedb, and take n_ar and n_bs as the crossover points of the two paths respectively.

[0105] Step S1332 - 82: Divide the node sequences of the two parent paths into a front segment sequence and a rear segment sequence at the crossover point, exchange the rear segment sequences to generate two offspring paths, and verify whether the straight-line distance between adjacent nodes in the offspring paths is less than the maximum single-segment flight distance of the drone. If it exceeds, insert intermediate nodes or reselect the crossover point.

[0106] Suppose the crossover points are n_ar (from P_selecteda) and n_bs (from P_selectedb).

[0107] Divide the node sequence of P_selecteda at n_ar into a front segment sequence {n_a1, n_a2,..., n_ar} and a rear segment sequence {n_a(r + 1), n_a(r + 2),..., n_am}; divide the node sequence of P_selectedb at n_bs into a front segment sequence {n_b1, n_b2,..., n_bs} and a rear segment sequence {n_b(s + 1), n_b(s + 2),..., n_bn}.

[0108] The back-end sequences are swapped to generate two offspring paths, P_offspring1 and P_offspring2. The node sequence of P_offspring1 is {n_a1, n_a2, …, n_ar, n_b(s+1), n_b(s+2), …, n_bn}; the node sequence of P_offspring2 is {n_b1, n_b2, …, n_bs, n_a(r+1), n_a(r+2), …, n_am}.

[0109] Assume that the maximum single-segment flight distance of the drone is D_max. For the child path P_offspring1, check the straight-line distance between adjacent nodes. For nodes n_ai and n_a(i+1) (i ranges from 1 to r-1), nodes n_ar and n_b(s+1), and subsequent adjacent nodes, calculate the straight-line distance d = [(x_na(i+1)-x_nai)^2 + (y_na(i+1)-y_nai)^2]^(1 / 2) between them. If the straight-line distance d between any adjacent nodes is greater than D_max, an intermediate node needs to be inserted.

[0110] The method for inserting an intermediate node can be to determine the coordinates of a new node between the two nodes according to a set rule. For example, linear interpolation can be performed based on the coordinates of the two nodes. Let n_ai = (x_nai, y_nai) and n_a(i+1) = (x_na(i+1), y_na(i+1)). The coordinates of the newly inserted intermediate node (x_mid, y_mid) can be obtained by the formula x_mid = (x_nai + x_na(i+1)) / 2, y_mid = (y_nai + y_na(i+1)) / 2 (this is just a simple interpolation method. In actual application, a more appropriate method can be selected according to the specific situation).

[0111] If the condition that the straight-line distance between adjacent nodes is less than D_max is still not met after inserting the intermediate node, a new intersection point is selected and the above splitting, swapping and verification process is repeated. The same operation is also applied to the offspring path P_offspring2.

[0112] Step S1332-83: performing mutation processing on the node sequence of the child path, the mutation processing comprising: randomly selecting nodes in the child path that are neither starting points nor ending points, generating replacement nodes according to the terrain connectivity parameters in the dynamic constraint conditions, and the replacement nodes must satisfy that the straight-line distance difference with the previous and next nodes in the original path does not exceed a preset tolerance value.

[0113] For the offspring path P_offspring1, randomly select a non-starting and non-ending node from its node sequence {n_a1, n_a2, ..., n_ar, n_b(s+1), n_b(s+2), ..., n_bn}. Assume that the selected node is n_at(1 <t<r+n-s)。

[0114] The terrain connectivity parameters in the dynamic constraint conditions include average elevation parameters, elevation fluctuation parameters, etc. Assume that the terrain connectivity parameter set is TC = {h_avg, h_var, ...}.

[0115] Generate a replacement node based on these parameters. For example, the coordinates of the replacement node can be determined based on the average elevation parameter h_avg and the elevation fluctuation parameter h_var, combined with the general trend of the surrounding terrain. Assuming the coordinates of the original node n_at are (x_nat, y_nat), considering the continuity of the surrounding terrain, the coordinates of the replacement node n_at' are generated according to the set rules (x_nat', y_nat').

[0116] After generating the replacement node, it is necessary to verify whether the straight-line distance difference between it and the previous and next nodes in the original path does not exceed the preset tolerance value. Calculate the straight-line distance d1 = [(x_nat-x_na(t-1))^2 + (y_nat-y_na(t-1))^2]^(1 / 2) between the original node n_at and the previous node n_a(t-1). Calculate the straight-line distance d2 = [(x_nat'-x_na(t-1))^2 + (y_nat'-y_na(t-1))^2]^(1 / 2) between the replacement node n_at' and the previous node n_a(t-1). Calculate the straight-line distance d3=[(x_nat-x_na(t+1))^2+(y_nat-y_na(t+1))^2]^(1 / 2) between the original node n_at and the next node n_a(t+1), and calculate the straight-line distance d4=[(x_nat'-x_na(t+1))^2+(y_nat'-y_na(t+1))^2]^(1 / 2) between the replacement node n_at' and the next node n_a(t+1).

[0117] If |d2-d1|≤Tolerance and |d4-d3|≤Tolerance, then accept the alternative node; otherwise, regenerate the alternative node until the conditions are met. The same mutation process is applied to the offspring path P_offspring2.

[0118] Step S1332 - 84: Constrained adjustment is performed on the mutated offspring path according to the meteorological constraint parameter set in the dynamic constraint conditions. The constrained adjustment includes: If the wind direction intensity parameter of the meteorological period corresponding to a node in the offspring path exceeds the wind resistance threshold of the unmanned aerial vehicle, the node coordinates are adjusted according to the included angle between the wind direction and the flight direction, so that the cosine value of the adjusted included angle is greater than the preset safety value.

[0119] The meteorological constraint parameter set includes the wind direction intensity parameter $W_j$ for each meteorological period. Let the wind resistance threshold of the unmanned aerial vehicle be $W\_threshold$, and the preset safety value be $Cos\_safe$.

[0120] For the mutated offspring path $P\_offspring1$, traverse its node sequence. Assume that the meteorological period corresponding to node $n\_au$ is $Tk$, and the wind direction intensity parameter for this period is $W\_k$. If $W\_k > W\_threshold$, then the coordinates of node $n\_au$ need to be adjusted.

[0121] Given that the flight direction vector of node $n\_au$ is $F\_u=(x\_Fu, y\_Fu)$, and the wind direction vector for this meteorological period is $W\_k=(x\_Wk, y\_Wk)$. According to the vector included angle formula $\cos\theta=(x\_Fu * x\_Wk + y\_Fu * y\_Wk) / ((x\_Fu^2 + y\_Fu^²)^(1 / 2) * (x\_Wk^2 + y\_Wk^2)^(1 / 2))$, calculate the current cosine value of the included angle $\cos\theta$.

[0122] If $\cos\theta < Cos\_safe$, then the coordinates $(x\_nau, y\_nau)$ of node $n\_au$ are adjusted according to the set rules. For example, the node $n\_au$ can be moved a certain distance along the direction perpendicular to the wind direction according to the wind direction vector $W\_k$. Let the moving distance be $\Delta d$, and the vector perpendicular to the wind direction can be obtained by rotating the wind direction vector $W\_k$ counterclockwise by 90 degrees. Let the rotated vector be $V\_rotate = (-y\_Wk, x\_Wk)$. The adjusted node coordinates $(x\_nau', y\_nau')=(x\_nau+\Delta d*(-y\_Wk) / (x\_Wk^2 + y\_Wk^2)^(1 / 2), y\_nau+\Delta d*x\_Wk / (x\_Wk^2 + y\_Wk^2)^(1 / 2))$.

[0123] Recalculate the cosine of the angle between the adjusted node's flight direction and the wind direction. If the adjusted cosine of the angle is greater than Cos_safe, the node is adjusted. If it is still less than Cos_safe, continue adjusting the movement distance Δd and recalculate the adjusted coordinates and cosine of the angle until the cosine of the angle is greater than Cos_safe. The same constraint adjustment operation is applied to all nodes in the child path P_offspring2 that meet the wind direction intensity parameter exceeding the drone's wind resistance threshold.

[0124] Step S1332-85: Merge the constraint-adjusted child paths with the screened candidate path set to generate an intermediate path set, and delete invalid paths in the intermediate path set whose path obstacle contact parameters are greater than zero.

[0125] The offspring paths P_offspring1 and P_offspring2 that have undergone mutation and constraint adjustment are merged with the screened candidate path set P_selected to obtain the intermediate path set P_intermediate.

[0126] For each path in the intermediate path set P_intermediate, its path obstacle contact parameters are recalculated. Taking a path P_temp as an example, its node sequence is {n_temp1, n_temp2, …, n_tempu}, and the terrain obstacle area coordinate set is O = {o1, o2, …, on}.

[0127] Traverse the node sequence of the path P_temp and, for each node n_tempi, calculate the distance d(n_tempi, oi) = [(x_n_tempi - x_oi)^2 + (y_n_tempi - y_oi)^2]^(1 / 2) between it and each coordinate oi within the terrain obstacle area. Compare this distance with the preset distance threshold D_threshold, which represents the extent of the obstacle area. If any node n_tempi exists such that d(n_tempi, oi) is less than or equal to D_threshold, then the path obstacle contact parameter N_obs_temp for that path is greater than zero, and the path is deemed invalid and removed from the intermediate path set P_intermediate.

[0128] After the above operations, only valid paths with zero path obstacle contact parameters are retained in the intermediate path set P_intermediate.

[0129] Step S1332-86: Repeat the step of randomly selecting two candidate paths from the filtered candidate path set until the number of paths in the intermediate path set reaches a preset size, and use the intermediate path set as the updated path set.

[0130] Set a default size value Size_threshold.

[0131] From the filtered candidate path set P_selected, two candidate paths are randomly selected as parent paths. According to the above steps of selecting intersections, generating child paths, mutation processing, constraint adjustment, and merging and deleting invalid paths, new paths are continuously generated and added to the intermediate path set P_intermediate.

[0132] After each new path is generated and processed, the number of paths in the intermediate path set P_intermediate is checked. If the number of paths is less than Size_threshold, the above operation is continued by randomly selecting two candidate paths from the filtered candidate path set P_selected. If the number of paths reaches or exceeds Size_threshold, the operation is stopped and the intermediate path set P_intermediate at this time is used as the updated path set P_updated.

[0133] This updated path set P_updated is a relatively better path set after genetic operation processing, taking into account static and dynamic constraints, and provides a better candidate path basis for the subsequent generation of global path solutions.

[0134] Step S1333: Repeat step S1332 until the iteration termination condition is reached, and then select the candidate path with the highest optimization score from the final path set as the global path solution.

[0135] After completing step S1332 and generating the updated path set P_updated, the iterative process begins. The iteration termination criteria can be set, such as reaching a preset maximum number of iterations (Max_iter), or when the optimization score of the candidate paths in the path set P_updated improves by less than a minimum threshold (Improve_threshold) over multiple consecutive iterations. The candidate path with the highest optimization score is then selected from the final path set P_updated_new as the global path solution. This iterative approach continuously optimizes the path set, resulting in a global path solution that is as optimal as possible while comprehensively considering static and dynamic constraints, meeting the requirements of drone swarm logistics distribution path planning.

[0136] Step S140: verifying the global path solution and the dynamic adjustment solution to generate a verification result set.

[0137] In the scenario of this embodiment, verification processing is performed based on the generated global path solution and dynamic adjustment solution.

[0138] Step S141: simulating the trajectory data of the UAV cluster flying according to the global path plan, and generating simulated interference parameters in combination with the meteorological constraint parameter set.

[0139] Assume that the sequence of nodes in the UAV flight path determined by the global path solution is {G1, G2, …, Gv}, and each node Gi has coordinates (x_Gi, y_Gi).

[0140] The simulated drone swarm flies along this path, recording relevant data at each node to generate trajectory data. For example, the position coordinates, flight speed, flight direction, and other information of each node are recorded to form trajectory data T_data = {(x_G1, y_G1, V1, F1), (x_G2, y_G2, V2, F2), …, (x_Gv, y_Gv, Vv, Fv)}, where Vi represents the flight speed at node Gi and Fi represents the flight direction vector at node Gi.

[0141] The meteorological constraint parameter set includes parameters such as wind direction intensity parameter W_j, rainfall range parameter R_cum_j, rainfall duration parameter t_r_j, temperature change parameter t_var_j, maximum temperature t_max_j and minimum temperature t_min_j for each meteorological period.

[0142] Combined with the set of meteorological constraint parameters, simulated interference parameters are generated. For example, for the wind direction intensity parameter W_j, the wind direction intensity parameter W_j for each node Gi is obtained based on the meteorological period Tj in that period. Assuming that the influence function of the wind direction intensity parameter on UAV flight is f_w(W_j), this function is used to calculate the interference value I_wi = f_w(W_j) of the wind direction on UAV flight at node Gi.

[0143] For the rainfall range parameter R_cum_j, assuming that the impact function of rainfall on UAV flight is f_r(R_cum_j), according to the rainfall range parameter R_cum_j in the meteorological period where the node Gi is located, the interference value of rainfall on UAV flight I_ri=f_r(R_cum_j) is calculated.

[0144] Similarly, for other meteorological constraint parameters, the interference values ​​on UAV flight are calculated according to the corresponding influence functions, such as the interference value I_tvi corresponding to the temperature change parameter, the interference value I_tmaxi corresponding to the maximum temperature, and the interference value I_tmini corresponding to the minimum temperature.

[0145] These interference values ​​are combined to form the simulated interference parameters I = {I_wi, I_ri, I_tvi, I_tmaxi, I_tmini, …}, which are used for subsequent evaluation of the UAV’s flight trajectory stability and energy consumption.

[0146] Step S142: superimposing and analyzing the trajectory data and the simulated interference parameters to generate trajectory stability parameters and energy consumption estimation parameters.

[0147] For trajectory data T_data={(x_G1, y_G1, V1, F1), (x_G2, y_G2, V2, F2),…, (x_Gv, y_Gv, Vv, Fv)} and simulated interference parameters I={I_wi, I_ri, I_tvi, I_tmaxi, I_tmini,…}.

[0148] First, we analyze the trajectory stability parameters. Considering the impact of wind interference on trajectory stability, we assume that wind interference can cause the drone's flight direction to deviate. At node Gi, based on the wind interference value I_wi and the flight direction vector Fi, we calculate the flight direction deviation ΔFi caused by the wind interference through a predefined vector operation. For example, let the wind direction vector be W_j = (x_Wj, y_Wj). Based on the wind intensity parameter W_j and the flight direction vector Fi, we calculate the additional vector V_add generated by the wind interference according to the principle of vector synthesis. Then, we convert V_add into the flight direction deviation ΔFi using a predefined proportional relationship.

[0149] Similarly, considering the impact of other simulated interference parameters such as rainfall and temperature on flight direction and speed, the corresponding deviations are calculated. These deviations are combined and, using a predefined statistical method, such as calculating the root mean square (RMS) value of all deviations, the trajectory stability parameter S is obtained. The formula for calculating the RMS value is assumed to be S = [(∑(ΔFi^2+ΔVi^2+…)) / v]^(1 / 2), where ΔVi represents the speed deviation due to various interferences and v is the number of path nodes.

[0150] For energy consumption estimation, consider the energy consumed by the drone to overcome various interferences during flight. Assume that the drone's energy consumption model is related to factors such as flight speed and the interference force it encounters. For node Gi, the energy consumption value E_i at that node is calculated using the energy consumption model function E(Vi, I_wi, I_ri, ...) based on simulated interference parameters such as flight speed Vi, wind direction interference value I_wi, and rainfall interference value I_ri.

[0151] The energy consumption values ​​at all nodes are accumulated to obtain the energy consumption estimation parameter E_total = ∑E_i (where i ranges from 1 to v). This energy consumption estimation parameter reflects the approximate energy consumption required for the drone to fly according to the global path plan while considering weather interference.

[0152] Step S143: generating an obstacle avoidance verification trajectory according to the obstacle avoidance path parameters in the dynamic adjustment scheme, and performing conflict detection processing on the obstacle avoidance verification trajectory and the terrain constraint parameter set to generate an obstacle avoidance success rate parameter.

[0153] The obstacle avoidance path parameters in the dynamic adjustment scheme include bypass distance parameters, priority parameters, etc. Assume that the obstacle avoidance path parameters are expressed as OBP = {D_bypass, P_priority, ...}, where D_bypass represents the bypass distance parameter and P_priority represents the priority parameter.

[0154] Generate an obstacle avoidance verification trajectory based on the obstacle avoidance path parameters. For example, when a terrain obstacle area is detected, the path nodes that bypass the obstacle area are determined according to the set rules based on the bypass distance parameter D_bypass. Assuming that the boundary of the obstacle area is represented by a series of coordinate points O_boundary = {o_b1, o_b2, …, o_bw}, based on the bypass distance parameter D_bypass, starting from a position a certain distance D_bypass away from the obstacle area boundary, a path node sequence {OB1, OB2, …, OBx} is generated in a set direction (such as clockwise or counterclockwise) to bypass the obstacle area, forming the obstacle avoidance verification trajectory T_OB = {(x_OB1, y_OB1), (x_OB2, y_OB2), …, (x_OBx, y_OBx)}.

[0155] The terrain constraint parameter set includes parameters such as the average elevation parameter h_avg, the elevation fluctuation parameter h_var, the maximum elevation difference parameter h_diff, and the terrain obstacle area coordinate set O={o1, o2,…, on}.

[0156] The obstacle avoidance verification trajectory T_OB is collided with the terrain constraint parameter set. For each node OBi = (x_OBi, y_OBi) in the obstacle avoidance verification trajectory T_OB, the distance d(OBi, oi) = [(x_OBi - x_oi)^2 + (y_OBi - y_oi)^2]^(1 / 2) between it and each coordinate oi in the terrain obstacle region coordinate set O is calculated and compared with the preset distance threshold D_threshold representing the range of the obstacle region.

[0157] Count the number of nodes \(N_{collision}\) in the statistical obstacle avoidance verification trajectory \(T_{OB}\) that conflict with the terrain obstacle area (i.e., \(d(OB_i, o_i)\leq D_{threshold}\)), and the total number of nodes \(N_{total}\) of the obstacle avoidance verification trajectory \(T_{OB}\).

[0158] The obstacle avoidance success rate parameter \(Success\_rate\) is calculated by the formula \(Success\_rate=(N_{total}-N_{collision}) / N_{total}\). This obstacle avoidance success rate parameter reflects the ability of the obstacle avoidance path in the dynamic adjustment scheme to avoid obstacles under actual terrain conditions.

[0159] Step S144: Integrate the trajectory stability parameter, the energy consumption estimation parameter, and the obstacle avoidance success rate parameter into the verification result set.

[0160] Integrate the previously calculated trajectory stability parameter \(S\), energy consumption estimation parameter \(E_{total}\), and obstacle avoidance success rate parameter \(Success\_rate\) together to form the verification result set \(V = \{S, E_{total}, Success\_rate\}\). This verification result set comprehensively reflects the performance of the global path scheme and the dynamic adjustment scheme under considerations of factors such as meteorology and terrain, providing a basis for subsequent judgment on whether the scheme meets the execution conditions.

[0161] Step S150: When the verification result set meets the execution conditions, send the global path scheme and the dynamic adjustment scheme to the UAV cluster for execution.

[0162] In the scenario of this embodiment, it is judged whether to send the scheme to the UAV cluster for execution based on the verification result set.

[0163] Step S151: If the trajectory stability parameter is less than the preset first stability threshold and the energy consumption estimation parameter is less than the preset energy consumption threshold, it is determined that the first execution condition is met.

[0164] Let the preset first stability threshold be \(S_{threshold1}\), and the preset energy consumption threshold be \(E_{threshold}\).

[0165] For the trajectory stability parameter \(S\) and the energy consumption estimation parameter \(E_{total}\) in the verification result set \(V = \{S, E_{total}, Success\_rate\}\). If \(S < S_{threshold1}\) and \(E_{total} < E_{threshold}\), it is determined that the first execution condition is met. This means that under the consideration of meteorological interference, the trajectory stability of the UAV flying according to the global path scheme is within an acceptable range, and the energy consumption estimation is also within the allowed threshold. From these two aspects, it preliminarily shows that the global path scheme has a certain feasibility.

[0166] Step S152: If the obstacle avoidance success rate parameter is greater than a preset success rate threshold and the matching degree between the weather adaptation parameter in the dynamic adjustment scheme and the weather constraint parameter set is greater than a preset matching threshold, it is determined that the second execution condition is met.

[0167] Assume that the preset success rate threshold is Success_threshold and the preset matching threshold is Match_threshold.

[0168] For the obstacle avoidance success rate parameter Success_rate in the verification result set V, if Success_rate>Success_threshold, it means that the obstacle avoidance path in the dynamic adjustment scheme has a high obstacle avoidance success rate under actual terrain conditions and can effectively avoid terrain obstacles.

[0169] The meteorological adaptation parameters in the dynamic adjustment scheme include, for example, altitude adjustment parameters and heading offset parameters. Assume that the meteorological adaptation parameters are represented by MAP={H_adjust, C_offset, …}, and the meteorological constraint parameter set is MC={W_j, R_cum_j, t_r_j, t_var_j, t_max_j, t_min_j, …}.

[0170] Calculate the degree of match between the weather adaptation parameters and the set of weather constraint parameters. This can be done by designing a matching function, Match(MAP, MC), which considers various relationships between weather adaptation parameters and weather constraint parameters, such as the relationship between altitude adjustment parameters and weather factors such as temperature and air pressure, and the relationship between heading offset parameters and weather factors such as wind direction and intensity. Assume that the matching degree calculated by this matching function is Match_value.

[0171] If Match_value>Match_threshold, the second execution condition is determined to be satisfied. This indicates that the dynamic adjustment scheme can better adapt to changes in meteorological conditions and has a good ability to cope with the influence of meteorological factors.

[0172] Step S153: When the first execution condition and the second execution condition are met at the same time, the global path plan is converted into a waypoint instruction set, and the dynamic adjustment plan is converted into a real-time correction instruction set, which is sent to the drone cluster through the communication module.

[0173] When the first execution condition and the second execution condition are met at the same time, it means that the global path plan and the dynamic adjustment plan have good feasibility and adaptability after comprehensively considering factors such as weather and terrain, and can be sent to the drone cluster for execution.

[0174] For a global path plan, the sequence of nodes in the drone's flight path is determined as {G1, G2, …, Gv}, where each node Gi has coordinates (x_Gi, y_Gi). The global path plan is converted into a waypoint instruction set. For example, the coordinates of each node (x_Gi, y_Gi) are converted into waypoint coordinate instructions according to a predetermined encoding rule. At the same time, an arrival time tag T_i is added to each waypoint, indicating the estimated arrival time of the drone at that waypoint. Assume that the encoding rule converts the coordinate values ​​into a predetermined data format, such as converting x_Gi and y_Gi into binary form, and then combining them into an instruction code of a specific length. The resulting waypoint instruction set WP = {(Code(x_G1, y_G1), T1), (Code(x_G2, y_G2), T2), …, (Code(x_Gv, y_Gv), Tv)}, where Code(x, y) represents the instruction code after encoding the coordinates (x, y).

[0175] For dynamic adjustment schemes, information such as weather adaptation parameters and obstacle avoidance path parameters needs to be converted into a set of real-time correction instructions. For example, consider the altitude adjustment parameter H_adjust within the weather adaptation parameters. If the altitude adjustment rule states that when the wind direction strength parameter W_j exceeds a certain threshold, the drone must ascend by a certain altitude ΔH, then this rule can be converted into a conditional trigger instruction. Let's assume this conditional trigger instruction is CT1 = {Trigger_condition1, Action1}, where Trigger_condition1 is "W_j > W_threshold1" and Action1 is "increase altitude by ΔH."

[0176] Obstacle avoidance path parameters such as the bypass distance parameter D_bypass and the priority parameter P_priority are similarly converted into conditional trigger instructions. For example, let the obstacle avoidance condition trigger instruction be CT2 = {Trigger_condition2, Action2}, where Trigger_condition2 is "a terrain obstacle area is detected and the priority P_priority meets a certain condition" and Action2 is "bypass the obstacle area according to the bypass distance parameter D_bypass."

[0177] These conditional trigger instructions are combined together to form a real-time correction instruction set RM = {CT1, CT2, ...}.

[0178] Finally, the waypoint instruction set WP and the real-time correction instruction set RM are sent to the UAV cluster through the communication module. After receiving these instructions, the UAV cluster can fly according to the global path plan and make dynamic adjustments according to the real-time correction instruction set during the flight to adapt to environmental changes such as weather and terrain and complete logistics distribution tasks.

[0179] Figure 2 A schematic diagram illustrates exemplary hardware and software components of a drone swarm-based logistics delivery path planning system 100 that can implement the concepts of the present application, as provided in some embodiments of the present application. For example, the processor 120 can be used in the drone swarm-based logistics delivery path planning system 100 to perform the functions described in the present application.

[0180] The drone swarm-based logistics distribution path planning system 100 can be a general-purpose server or a special-purpose server, both of which can be used to implement the drone swarm-based logistics distribution path planning method of this application. Although only one server is shown in this application, for convenience, the functions described in this application can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0181] For example, the logistics distribution path planning system 100 based on the drone swarm may include a network port 110 connected to the network, one or more processors 120 for executing program instructions, a communication bus 130, and storage media 140 in different forms, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the logistics distribution path planning system 100 based on the drone swarm may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The method of the present application can be implemented according to these program instructions. The logistics distribution path planning system 100 based on the drone swarm also includes an I / O interface 150 between the computer and other input and output devices.

[0182] For ease of explanation, only one processor is described in the drone swarm-based logistics distribution path planning system 100. However, it should be noted that the drone swarm-based logistics distribution path planning system 100 in this application may also include multiple processors, so the steps performed by one processor described in this application may also be performed jointly or individually by multiple processors. For example, if the processor of the drone swarm-based logistics distribution path planning system 100 executes step A and step B, it should be understood that step A and step B may also be performed jointly by two different processors or individually in one processor. For example, the first processor executes step A, the second processor executes step B, or the first processor and the second processor execute steps A and B together.

[0183] In addition, an embodiment of the present invention also provides a readable storage medium, in which computer-executable instructions are preset. When the processor executes the computer-executable instructions, the above-mentioned logistics distribution path planning method based on drone clusters is implemented.

[0184] It should be noted that in order to simplify the description of the present invention and thus help understand one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, multiple features are sometimes combined into one embodiment, figure or description thereof.

Claims

1. A logistics distribution path planning method based on drone swarm, characterized by: The method comprises: Obtaining logistics task data and environmental monitoring data for the target area, wherein the logistics task data includes a set of delivery starting points, delivery destinations, and delivery task priorities, and the environmental monitoring data includes meteorological data and terrain data; Performing feature extraction on the logistics task data and the environmental monitoring data to generate a path planning feature set and an environmental constraint feature set; Performing path optimization processing based on the path planning feature set and the environmental constraint feature set to generate a global path plan and a dynamic adjustment plan for the UAV cluster; Performing verification processing on the global path solution and the dynamic adjustment solution to generate a verification result set; When the verification result set meets the execution condition, the global path plan and the dynamic adjustment plan are sent to the drone cluster for execution; The feature extraction of the logistics task data and the environmental monitoring data to generate a path planning feature set and an environmental constraint feature set includes: Segmenting the delivery starting point and the delivery destination set according to the delivery task priority to generate multiple sub-path segments and corresponding time constraint parameters as the path planning feature set, each sub-path segment including a start coordinate and an end coordinate; Performing regional impact analysis on the meteorological data to generate a set of meteorological constraint parameters; Performing terrain connectivity analysis on the terrain data to generate a set of terrain constraint parameters; Combine the meteorological constraint parameter set and the terrain constraint parameter set to generate the environmental constraint feature set; The verifying the global path solution and the dynamic adjustment solution to generate a verification result set includes: Simulating trajectory data of a UAV cluster flying according to the global path plan, and generating simulated interference parameters in combination with the meteorological constraint parameter set; Superimposing and analyzing the trajectory data and the simulated interference parameters to generate trajectory stability parameters and energy consumption estimation parameters; generating an obstacle avoidance verification trajectory according to the obstacle avoidance path parameters in the dynamic adjustment scheme, and performing conflict detection processing on the obstacle avoidance verification trajectory and the terrain constraint parameter set to generate an obstacle avoidance success rate parameter; The trajectory stability parameter, the energy consumption estimation parameter, and the obstacle avoidance success rate parameter are integrated into the verification result set.

2. The logistics distribution path planning method based on drone swarm according to claim 1 is characterized in that: The performing of regional impact analysis on the meteorological data to generate a set of meteorological constraint parameters includes: Dividing the meteorological data into multiple meteorological periods according to the time dimension, each meteorological period includes a wind speed sequence, a rainfall sequence and a temperature sequence; Peak extraction is performed on the wind speed sequence for each meteorological period to generate maximum and average wind speed values, and the wind direction intensity parameter for each period is calculated based on the UAV aerodynamic model. The wind direction intensity parameter is defined as the cosine value of the absolute value of the angle between the UAV flight direction and the wind direction; Performing cumulative calculation processing on the rainfall sequence in each meteorological period to generate rainfall range parameters and rainfall duration parameters; Performing difference calculation on the temperature sequence of each meteorological period to generate temperature change parameters and extreme temperature parameters; The wind direction intensity parameter, the rainfall range parameter, the rainfall duration parameter, the temperature change parameter and the extreme temperature parameter are integrated into the meteorological constraint parameter set in chronological order.

3. The logistics distribution path planning method based on drone swarm according to claim 1 is characterized in that: The performing terrain connectivity analysis on the terrain data to generate a set of terrain constraint parameters includes: Converting the terrain data into an elevation grid map, and performing region division processing on the elevation grid map to generate a plurality of connected regions; Performing statistical calculations on the elevation grid map of each connected area to generate average elevation parameters, elevation fluctuation parameters, and maximum elevation difference parameters; Perform mutation point identification processing on the elevation grid map according to a preset elevation mutation threshold to generate a coordinate set of terrain obstacle areas; The average elevation parameter, the elevation fluctuation parameter, the maximum elevation difference parameter, and the terrain obstacle area coordinate set are integrated into the terrain constraint parameter set.

4. The logistics distribution path planning method based on drone swarm according to claim 1 is characterized in that: The performing of path optimization processing based on the path planning feature set and the environmental constraint feature set to generate a global path plan and a dynamic adjustment plan for the UAV cluster includes: Constructing static constraints based on the sub-path segments and time constraint parameters in the path planning feature set, and dynamically correcting the preset cruising speed of the UAV in combination with the wind speed parameters in the real-time meteorological data; generating a dynamic constraint condition according to the revised preset cruising speed, the set of meteorological constraint parameters, and the set of terrain constraint parameters; Calling a path optimization algorithm to iteratively solve the static constraint conditions and the dynamic constraint conditions to generate the global path solution; The dynamic adjustment scheme is generated based on the real-time change data of the dynamic constraint conditions, and the dynamic adjustment scheme includes obstacle avoidance path parameters and weather adaptation parameters.

5. The logistics distribution path planning method based on drone swarm according to claim 4 is characterized in that: Generating the dynamic adjustment scheme based on the real-time changing data of the dynamic constraint condition includes: Monitoring real-time change data of wind direction intensity parameters and rainfall range parameters in the meteorological constraint parameter set; When it is detected that the wind direction intensity parameter exceeds a preset wind speed threshold or the rainfall range parameter is extended, generating a meteorological adaptation parameter including an altitude adjustment parameter and a heading offset parameter; Monitoring update data of a terrain obstacle region coordinate set in the terrain constraint parameter set; When a new obstacle area is detected or the range of the obstacle area changes, obstacle avoidance path parameters including detour distance parameters and priority parameters are generated; The weather adaptation parameters and the obstacle avoidance path parameters are integrated into the dynamic adjustment scheme.

6. The logistics distribution path planning method based on drone swarm according to claim 1 is characterized in that: When the verification result set satisfies the execution condition, the global path plan and the dynamic adjustment plan are sent to the drone cluster for execution, including: If the trajectory stability parameter is less than a preset first stability threshold and the energy consumption estimation parameter is less than a preset energy consumption threshold, determining that a first execution condition is met; If the obstacle avoidance success rate parameter is greater than a preset success rate threshold and the matching degree between the weather adaptation parameter in the dynamic adjustment scheme and the weather constraint parameter set is greater than a preset matching threshold, it is determined that the second execution condition is met; When the first execution condition and the second execution condition are simultaneously met, the global path plan is converted into a waypoint instruction set, and the dynamic adjustment plan is converted into a real-time correction instruction set, which is sent to the UAV cluster through the communication module; The converting of the global path plan into a waypoint instruction set and the converting of the dynamic adjustment plan into a real-time correction instruction set includes: Encoding a sequence of nodes in the global path solution into waypoint coordinates and arrival time tags in chronological order; Encoding the weather adaptation parameters and obstacle avoidance path parameters in the dynamic adjustment scheme into conditional trigger instructions, each conditional trigger instruction including a trigger threshold and execution parameters; Distributing the waypoint coordinates, the arrival time tag, and the conditional trigger instruction to the control unit of each drone; The unexecuted waypoint coordinates and the conditional trigger instructions are dynamically adjusted according to the status feedback data of each drone.

7. The logistics distribution path planning method based on drone swarm according to claim 4 is characterized in that: The calling of the path optimization algorithm to iteratively solve the static constraint conditions and the dynamic constraint conditions to generate the global path solution includes: Initialize a path set containing multiple candidate paths, each candidate path contains a node sequence and a cost parameter; Calculating an optimization score for each candidate path according to the static constraint conditions, screening the candidate paths according to the dynamic constraint conditions, performing genetic operations on the screened candidate paths, and generating an updated path set; Repeat the steps of calculating the optimization score of each candidate path according to the static constraint conditions, screening the candidate paths according to the dynamic constraint conditions, performing genetic operation processing on the screened candidate paths, and generating an updated path set until the iteration termination condition is reached, and then selecting the candidate path with the highest optimization score from the final path set as the global path solution.

8. A logistics distribution path planning system based on drone swarms, characterized by: It includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the logistics distribution path planning method based on drone clusters as described in any one of claims 1 to 7 above.

Images