Vehicle-unmanned aerial vehicle cooperative distribution path optimization method based on improved GA

Abstract

The invention relates to the technical field of logistics path planning, in particular to a vehicle-unmanned aerial vehicle cooperative distribution path optimization method based on an improved GA, and the method comprises the steps: constructing a vehicle and unmanned aerial vehicle cooperative distribution path optimization model, and taking the minimization of the total driving distance as an optimization target; the vehicle path and the unmanned aerial vehicle path are solved and optimized based on the improved genetic algorithm, unmanned aerial vehicle task allocation is performed after corresponding constraint conditions are met, and the quality of the solution is ensured through the convergence characteristic of the genetic algorithm. Specifically, the advantages of the unmanned aerial vehicle in the aspect of service centralized nodes are exerted, and the vehicle focuses on distribution of dispersed nodes, so that a good synergistic effect is formed. According to experimental verification, reasonable algorithm parameters are set, the method has important application value in an actual logistics distribution scene, the distribution cost can be remarkably reduced, and the operation efficiency can be improved.

Description

Technical Field

[0001] This invention relates to the field of logistics route planning technology, and specifically to a vehicle-drone collaborative delivery route optimization method based on improved GA. Background Technology

[0002] Special needs exist in real-world logistics scenarios: some customer locations cannot be reached directly by traditional vehicles due to geographical limitations (such as mountainous areas, remote areas, or traffic-controlled areas), while drones can effectively supplement the shortcomings of traditional delivery methods due to their flexibility and accessibility.

[0003] The core issue lies in how to coordinate a delivery vehicle carrying drones to efficiently plan the launch, service, and retrieval processes of the drones while completing regular customer point deliveries. The vehicle departs from the distribution center, visiting all reachable customer points along an optimized route; drones are released at appropriate locations to serve special customer points inaccessible by the vehicle; after completing their delivery tasks, the drones rendezvous with the vehicle at suitable locations along its path. This collaborative delivery model faces multiple complex constraints: drones have a maximum flight distance limit and must complete all assigned tasks within their range; drones can only be launched from vehicle docking points and must return to unvisited nodes along the vehicle's path; each customer point can only be visited once, and overall delivery efficiency must be optimized.

[0004] Through mathematical modeling and algorithm design, a collaborative delivery solution that minimizes the total travel distance is found, thereby reducing operating costs, improving delivery efficiency, and providing theoretical support and technical solutions for intelligent logistics systems. Last-mile delivery, as the most costly and least efficient link in the logistics chain, is often constrained by factors such as traffic congestion, dispersed customers, or poor accessibility in remote areas, thus limiting the overall service level. Summary of the Invention

[0005] The purpose of this invention is to provide a vehicle-drone collaborative delivery route optimization method based on improved GA, which aims to solve the problem of comprehensively utilizing the long-distance transportation capacity of vehicles and the flexible shuttle advantage of drones to improve overall delivery efficiency and reduce costs.

[0006] To achieve the above objectives, this invention provides a vehicle-drone collaborative delivery route optimization method based on improved GA, comprising the following steps:

[0007] Step 1: Architect a path optimization model, in which there is only one delivery vehicle and one drone;

[0008] Step 2: Define the customer node type and characteristics in the path optimization model;

[0009] Step 3: Set the optimization objective function and establish corresponding constraints;

[0010] Step 4: Solve the vehicle path and the UAV path based on the improved genetic algorithm, and output the final optimization result.

[0011] Optionally, in the path optimization model, the UAV is launched from the vehicle and can perform tasks multiple times at different points along the vehicle's path.

[0012] Optionally, the customer node types in step 2 include two categories: vehicle-accessible and drone-only accessible. Each type of node can only be served by the corresponding means of transportation, and each node must be visited once.

[0013] Optionally, the optimization objective in step 3 is to minimize the total travel distance, without considering time windows and dynamic environmental constraints; the total distance of a single drone flight is subject to an upper limit constraint, but it is assumed that the vehicle has sufficient range to meet the travel distance requirements of all delivery tasks.

[0014] Optionally, the optimization solution process in step 4 includes the following steps:

[0015] Step 4.1: Initialize parameter settings, including delivery network size parameters, drone performance parameters, and genetic algorithm running parameters;

[0016] Step 4.2: Initialize the vehicle path population;

[0017] Step 4.3: Evaluate the path cost for individual vehicles in the population;

[0018] Step 4.4: Assign drone missions to the optimal vehicle routes;

[0019] Step 4.5: Population selection, crossover, mutation, and population renewal;

[0020] Step 4.6: Perform 2-opt local optimization;

[0021] Step 4.7: Optimize the path for each drone mission;

[0022] Step 4.8: Output the final optimization results.

[0023] Optionally, during the execution of step 4.4, cluster analysis is first performed on the UAV nodes to determine whether a service cluster can be formed within the UAV's flight range; then the optimal launch and recovery point is determined.

[0024] Optionally, the execution process of step 4.7 includes the following steps:

[0025] The nearest neighbor heuristic algorithm is used to construct the drone visit sequence;

[0026] Starting from the launch point, select the nearest unvisited node in sequence until all cluster nodes are served;

[0027] Finally, plan the path from the last service node to the recycling point.

[0028] This invention provides a vehicle-drone collaborative delivery route optimization method based on an improved Genetic Algorithm (GA). By constructing a route optimization model for vehicle-drone collaborative delivery, with the goal of minimizing the total travel distance, an improved GA is used to optimize both vehicle and drone routes. Drone tasks are assigned after satisfying relevant constraints, and the convergence characteristics of the GA ensure the quality of the solution. Specifically, this method leverages the advantages of drones in serving centralized nodes, while vehicles focus on delivery to decentralized nodes, creating a good synergistic effect. Experiments have verified that, with reasonable algorithm parameters, this method has significant application value in real-world logistics delivery scenarios, significantly reducing delivery costs and improving operational efficiency. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a schematic diagram of the improved genetic algorithm process in the vehicle-drone collaborative delivery route optimization method based on improved GA of the present invention.

[0031] Figure 2 This is a schematic diagram of the optimized path for vehicles and drones according to an embodiment of the present invention.

[0032] Figure 3 This is a schematic diagram of the vehicle path distance convergence curve and the total distance cost convergence curve between the vehicle and the drone in an embodiment of the present invention. Detailed Implementation

[0033] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0034] The following is a description of the English abbreviations and terms used in this invention:

[0035] GA: Genetic Algorithm.

[0036] This invention provides a vehicle-drone collaborative delivery route optimization method based on improved GA, comprising the following steps:

[0037] Step 1: Architect a path optimization model, in which there is only one delivery vehicle and one drone;

[0038] Step 2: Define the customer node type and characteristics in the path optimization model;

[0039] Step 3: Set the optimization objective function and establish corresponding constraints;

[0040] Step 4: Solve the vehicle path and the UAV path based on the improved genetic algorithm, and output the final optimization result.

[0041] The following provides further explanation with reference to specific embodiments and execution processes:

[0042] The present invention makes the following assumptions regarding the improved vehicle-drone collaborative delivery route optimization method for GA:

[0043] 1. It consists of a delivery vehicle and a drone, which is launched from the vehicle and can perform multiple missions at different points along the vehicle's route;

[0044] 2. Customer points are clearly divided into two categories: vehicle-accessible and drone-only accessible. Each type of node can only be served by the corresponding means of transportation, and each node must be visited once.

[0045] 3. The objective is to minimize the total driving distance, without considering complex factors such as time windows and dynamic environments.

[0046] 4. The total distance of a single drone flight is subject to an upper limit, while vehicles are assumed to have sufficient range to meet the travel distance requirements of all delivery tasks.

[0047] The process of constructing the path optimization model is as follows:

[0048] Table 1 Symbol Explanation Table

[0049]

[0050] Objective function: Minimize total cost

[0051]

[0052] constraint:

[0053]

[0054]

[0055]

[0056]

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063]

[0064]

[0065]

[0066]

[0067]

[0068]

[0069] Formula (1) minimizes the weighted sum of the vehicle's travel distance and the drone's flight distance. Formula (2) indicates that each vehicle node must have an outgoing edge to ensure that the node is visited. Formula (3) indicates that each vehicle node must have an incoming edge to ensure that the node is visited. Formula (4) indicates that the warehouse must have an outgoing edge to the customer point to start the delivery trip, and must have an edge from the customer point back to the warehouse to complete the delivery trip. Formula (5) is used to eliminate sub-loop constraints to ensure that the path is a single continuous loop. Formula (6) indicates that the warehouse's sequence number is fixed at 0, serving as the starting point of the path, and the customer point's sequence number starts from 1 and increments. Formula (7) indicates that each drone node must be served by exactly one drone task. Formula (8) indicates that only active tasks can serve drone nodes. Formula (9) indicates that if a node is served, the task must be active. Formula (10) indicates that an active task must start from the launch point. Formula (11) indicates that an active task must return to the return point. Formula (12) indicates that the inflow to each intermediate node in the drone path equals the outflow to ensure path continuity. Formula (13) states that a node is served if and only if a path enters that node. Formula (14) states that the flight distance for each UAV mission cannot exceed the maximum limit. Formula (15) states that the return point must be visited after the launch point. Formula (16) states that each vehicle node can serve as the launch point for at most one mission. Formula (17) states that each vehicle node can serve as the return point for at most one mission.

[0070] The solution method based on genetic algorithm in step 4 is shown in the flowchart below. Figure 1 As shown, the key steps are as follows:

[0071] 1. Initialize parameters:

[0072] First, initialize the parameters required for the optimization algorithm:

[0073] (1) Set the delivery network scale parameters, including the coordinates of delivery points (divided into vehicle delivery points and drone delivery points), and the coordinates of delivery centers.

[0074] (2) Configure the performance parameters of the UAV and set the maximum flight distance of the UAV.

[0075] (3) Set the genetic algorithm running parameters, including population size, maximum number of iterations, crossover probability, and mutation probability.

[0076] 2. Initialize the vehicle path population

[0077] The initial vehicle path is generated by randomly arranging the nodes reachable by the vehicle. Each individual represents a complete vehicle visit sequence. For example, the individual [0,1,5,7,2,3,4,6,0] indicates that the vehicle starts from distribution center 0, passes through distribution points 1,5,7,2,3,4,6 in sequence, and finally returns to distribution center 0. A set number of individuals are randomly generated for the population size.

[0078] 3. Assess the individual vehicle route costs within the population.

[0079] Cost is evaluated for the vehicle path of each individual in the population. The cost is the distance traveled along the calculated path. For example, the cost of individual [0,1,5,7,2,3,4,6,0] is the distance traveled by the vehicle from distribution center 0, passing through distribution points 1,5,7,2,3,4,6 and finally returning to distribution center 0.

[0080] 4. Assign drone missions to the optimal vehicle routes.

[0081] Select the individual with the lowest cost from the vehicle path population and assign drone service tasks to the current optimal vehicle path:

[0082] 4.1 Cluster Analysis of Unmanned Aerial Vehicle Nodes

[0083] (1) Perform cluster analysis on the dedicated nodes of UAVs to determine whether a service cluster can be formed within the flight range of the UAV.

[0084] (2) An adaptive clustering method based on range constraints is adopted: when the estimated distance of the cluster formed by all UAV nodes does not exceed the maximum range of the UAV, it is considered as a cluster; otherwise, the node set is divided into two clusters on an average basis.

[0085] Cluster feasibility assessment formula:

[0086]

[0087] in, This is an estimate of the path length of the drone swarm S. This represents the maximum range of the drone.

[0088] 4.2 Determination of the optimal launch and recovery point

[0089] (1) Traverse all possible launch point locations along the vehicle path.

[0090] (2) For each launch point, find a suitable recovery point in the subsequent path nodes.

[0091] (3) Evaluate the collaborative delivery cost of each (launch point, recovery point) combination.

[0092] Collaborative cost assessment formula:

[0093]

[0094] in, Let r be the distance the vehicle travels from launch point l to recovery point r. drone service cluster Flight distance, This represents the cost coefficient for drones.

[0095] 5. Population selection, crossover, mutation, and population renewal.

[0096] The population is updated using a classic genetic algorithm, and the specific steps are as follows:

[0097] Selection: A roulette wheel selection method is used, where individuals with higher fitness have a higher probability of being selected.

[0098] Crossover: Using the sequential crossover (OX) method, superior gene sequences from the parent generation are preserved.

[0099] Mutation: A crossover mutation strategy is employed, randomly swapping two gene positions to maintain population diversity.

[0100] 6. 2-opt local optimization

[0101] The specific steps are as follows:

[0102] (1) Traverse all possible edge swap combinations in the vehicle path.

[0103] (2) For each pair of edges (i, i+1) and (j, j+1), try the path reversal operation.

[0104] (3) If the new path is cheaper, then accept the improvement.

[0105] Mathematical description of the 2-opt operation:

[0106] Given path ,for Generate a new path:

[0107]

[0108] Acceptance criteria: Path Cost less than path cost

[0109] 7. Drone Path Optimization Steps

[0110] Optimize the path for each drone mission:

[0111] (1) Constructing UAV visit sequences using the nearest neighbor heuristic algorithm

[0112] (2) Starting from the launch point, select the nearest unvisited node in sequence until all cluster nodes are served.

[0113] (3) Finally, plan the path from the last service node to the recycling point.

[0114] Drone path construction formula:

[0115]

[0116] in, As the launch point, For recycling points, For drone service node clusters, .

[0117] 8. Steps for outputting the optimal solution

[0118] Output the final optimization result:

[0119] (1) Output the optimal vehicle path sequence and UAV path sequence

[0120] (2) Output total delivery cost

[0121] By executing the above steps in an orderly manner, this invention achieves efficient optimization of the collaborative delivery route between vehicles and drones, significantly reducing delivery operation costs while ensuring service quality.

[0122] Furthermore, this invention includes simulation experiments for verification and explanation:

[0123] The experiment was conducted on a system with a Python 3.13 environment and a hardware configuration including an Intel i5-12600KF CPU, 32.0 GB of memory, and an NVIDIA GeForce RTX 4060 Ti GPU.

[0124] Key parameters are set as follows: For problem size, the total number of nodes is set to 15 (including the distribution center), and the maximum drone flight distance is 100 distance units; for algorithm parameters, the population size is set to 50, the number of generations is 100, and the crossover and mutation probabilities are set to 0.8 and 0.1 respectively, to balance the algorithm's global exploration and local exploitation capabilities; Parameter settings That is, the cost weight of the drone flight distance is 20% of the cost weight of the vehicle travel distance. Regarding the experimental data, randomly generated simulation data was used, including 1 logistics center, 10 vehicle-deliverable nodes, and 4 drone-only delivery nodes (nodes inaccessible by vehicles). The specific data distribution is shown in Table 2 below:

[0125] Table 2 Demand Points and Starting Point Data Information

[0126]

[0127] Through multiple experiments, a set of experimental results was randomly selected. Figure 2 The image shows the optimized path diagram for the vehicle and the drone. Figure 3 This represents the convergence curve for the vehicle path distance and the convergence curve for the total distance cost between the vehicle and the drone.

[0128] Experimental results show that the vehicle-drone collaborative delivery path optimization method based on improved GA of this invention can effectively solve complex logistics delivery optimization problems. The algorithm successfully found an efficient vehicle path with a total length of 292.25, and rationally allocated drone tasks, enabling the drone to start from node 14, serve four dedicated nodes, and return to the same node, with a flight distance of 98.32. The total operating cost was 311.91, demonstrating the algorithm's optimization capability in balancing vehicle and drone resources. The drone task allocation strategy fully leverages the advantages of drones in serving concentrated nodes, while vehicles focus on delivery to dispersed nodes, forming a good synergistic effect. The convergence characteristics of the genetic algorithm ensure the quality of the solution, proving that this invention has significant application value in real-world logistics delivery scenarios, significantly reducing delivery costs and improving operational efficiency.

[0129] The above description discloses only one or more preferred embodiments of the present invention, and should not be construed as limiting the scope of the present invention. Those skilled in the art will understand that all or part of the processes of the above embodiments can be implemented, and equivalent changes made in accordance with the claims of the present invention are still within the scope of the invention.

Claims

1. A method for optimizing vehicle-drone collaborative delivery routes based on improved GA, characterized in that, Includes the following steps: Step 1: Architect a path optimization model, in which there is only one delivery vehicle and one drone; Step 2: Define the customer node type and characteristics in the path optimization model; Step 3: Set the optimization objective function and establish corresponding constraints; Step 4: Solve the vehicle path and the UAV path based on the improved genetic algorithm, and output the final optimization result.

2. The vehicle-drone collaborative delivery route optimization method based on improved GA as described in claim 1, characterized in that, In the path optimization model, the UAV is launched from the vehicle and can perform tasks multiple times at different points along the vehicle's path.

3. The vehicle-drone collaborative delivery route optimization method based on improved GA as described in claim 2, characterized in that, The customer node types in step 2 include two categories: vehicle-accessible and drone-only accessible. Each type of node can only be served by the corresponding mode of transportation, and each node must be visited once.

4. The vehicle-drone collaborative delivery route optimization method based on improved GA as described in claim 3, characterized in that, The optimization objective in step 3 is to minimize the total travel distance, without considering time windows and dynamic environmental constraints; the total distance of a single drone flight has an upper limit constraint, but it is assumed that the vehicle has sufficient range to meet the travel distance requirements of all delivery tasks.

5. The vehicle-drone collaborative delivery route optimization method based on improved GA as described in claim 4, characterized in that, Step 4, the optimization solution execution process, includes the following steps: Step 4.1: Initialize parameter settings, including delivery network size parameters, drone performance parameters, and genetic algorithm running parameters; Step 4.2: Initialize the vehicle path population; Step 4.3: Evaluate the path cost for individual vehicles in the population; Step 4.4: Assign drone missions to the optimal vehicle routes; Step 4.5: Population selection, crossover, mutation, and population renewal; Step 4.6: Perform 2-opt local optimization; Step 4.7: Optimize the path for each drone mission; Step 4.8: Output the final optimization results.

6. The vehicle-drone collaborative delivery route optimization method based on improved GA as described in claim 5, characterized in that, During the execution of step 4.4, cluster analysis is first performed on the UAV nodes to determine whether a service cluster can be formed within the UAV's flight range; then the optimal launch and recovery point is determined.

7. The vehicle-drone collaborative delivery route optimization method based on improved GA as described in claim 6, characterized in that, The execution process of step 4.7 includes the following steps: The nearest neighbor heuristic algorithm is used to construct the drone visit sequence; Starting from the launch point, select the nearest unvisited node in sequence until all cluster nodes are served; Finally, plan the path from the last service node to the recycling point.

Images