Task planning and path optimization method and device for cooperation of multiple unmanned aerial vehicles and vehicles

Abstract

The invention discloses a task planning and path optimization method and device for cooperation of multiple unmanned aerial vehicles and vehicles. The method comprises the following steps: establishing an unmanned aerial vehicle and vehicle cooperation model; parking anchor points and a task point set corresponding to each parking anchor point are selected by using a K-means clustering algorithm; aiming at the task point set in the parking anchor point, acquiring an unmanned aerial vehicle flight route set; each route in the unmanned aerial vehicle flight route set is distributed to each unmanned aerial vehicle, and the unmanned aerial vehicle flight route in a single parking anchor point is obtained; a vehicle driving route is obtained according to all the parking anchor points, so that a global static optimal solution is obtained; and aiming at the real-time change of the speed of the unmanned aerial vehicle, adjusting the flight path of the unmanned aerial vehicle in a single anchor point through a tabu search algorithm, and outputting a local dynamic optimal solution. The advantage of cooperative work of the unmanned aerial vehicle and the vehicle is fully considered, and the system flexibility and the task response efficiency are improved under the constraint condition and the unmanned aerial vehicle speed change, so that the response delay of the system to the user request is greatly reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of cooperative operation between unmanned aerial vehicles and vehicles, and in particular to a method and device for task planning and path optimization for cooperation between multiple unmanned aerial vehicles and vehicles. Background Art

[0002] In recent years, with the rapid development of drone technology, the potential of drones in various application scenarios has gradually emerged. The lightweight and high flexibility of drones are gradually becoming a powerful supplement to traditional methods. At the same time, the collaborative operation of drones and vehicles has become a research hotspot in the field of intelligent transportation and data collection. Compared with the traditional single operation mode, the collaborative operation of drones and vehicles provides stronger adaptability and higher task execution efficiency.

[0003] Currently, many research teams have developed a variety of UAV-vehicle collaboration frameworks to achieve data collection and transportation tasks in complex environments. These UAVs need to frequently communicate and exchange data with vehicles when performing tasks to support large-scale real-time data collection applications. In a dynamically changing environment, UAVs and vehicles may face speed fluctuations and service demand changes every second. However, due to the dynamic changes caused by speed fluctuations, the path planning and task allocation of UAVs become extremely complex.

[0004] In existing research, the path planning and task allocation of UAVs are usually regarded as two independent parts. Although this method simplifies the planning process to a certain extent, it may lead to the performance of the overall solution not being optimal, especially in a dynamically changing environment. This method obviously cannot meet actual needs. Therefore, how to effectively deal with the dynamic changes caused by speed fluctuations and reduce the response delay problem of UAVs and vehicles collaborating to perform large-scale real-time data collection tasks has become a key issue that needs to be solved in this field.

[0005] The above contents are only used to assist in understanding the technical solution of the present invention and do not constitute an admission that the above contents are prior art. Summary of the invention

[0006] The purpose of the present invention is to overcome the above-mentioned defects and problems existing in the prior art, and to provide a method and device for task planning and path optimization of multi-UAV and vehicle collaboration, which fully considers the advantages of the collaborative work of UAVs and vehicles, and improves the flexibility of the method and the efficiency of task response under constraints and changes in UAV speed, thereby greatly reducing the response delay of the method to user requests.

[0007] To achieve the above objectives, the technical solution of the present invention is: a task planning and path optimization method for multi-UAV and vehicle collaboration, comprising:

[0008] Establish a UAV and vehicle collaboration model, which includes multiple mission points, multiple UAVs and at least one vehicle;

[0009] The K-means clustering algorithm is used to select the docking anchor points and the task point set corresponding to each docking anchor point; the UAV flight route set is obtained for the task point set within the docking anchor point; each route in the UAV flight route set is assigned to each UAV to obtain the UAV flight route within a single docking anchor point; the vehicle driving route is obtained based on all the docking anchor points, thereby obtaining the global static optimization solution of the UAV and vehicle collaboration model;

[0010] In response to the real-time changes in the UAV's speed, the UAV's flight path within a single anchor point is adjusted through the taboo search algorithm, and the local dynamic optimization solution of the UAV and vehicle collaboration model is output.

[0011] The method of selecting the docking anchor points and the task point set corresponding to each docking anchor point by using the K-means clustering algorithm includes:

[0012] Input the set of task points to be served, the set of anchor points to be selected, and the anchor point location of the current vehicle;

[0013] Randomly select K anchor points from the set of anchor points to be selected as the centroid;

[0014] For each task point to be served, assign it to the cluster set with the nearest centroid;

[0015] Calculate the total distance from all the task points to be served in each cluster set to the centroid, select an anchor point with the shortest distance as the new centroid, and continue to allocate the task points to be served;

[0016] After the initial clustering is completed, check whether the distance between the task point to be served in the cluster set and its centroid exceeds the maximum flight distance limit of the drone. If so, put the task point to be served into the task point set to be selected and reselect an anchor point for it.

[0017] If the anchor point where the vehicle is located is not selected as the centroid anchor point, it will be added to the centroid set, and according to the final centroid set, the task points to be served will be assigned one by one to the cluster set where the nearest centroid is located;

[0018] Add each centroid and its corresponding cluster set to the task point set corresponding to the anchor point.

[0019] The step of obtaining a UAV flight route set for a mission point set within a docking anchor point includes:

[0020] Input the anchor point position and the task point set corresponding to the anchor point;

[0021] Set a temporary task point set containing all unvisited task points;

[0022] A task point is randomly selected from the temporary task point set and added to the current route. The distance from the anchor point through the route and back to the anchor point is calculated, and it is determined whether the distance exceeds the maximum flight distance of the drone. If not, the task point is deleted from the temporary task point set and the task point is used as the last node position of the current route. If it exceeds, the task point is deleted from the route and the route is added to the drone flight route set.

[0023] After multiple iterations until the temporary mission point set is empty, the UAV flight route set at the anchor point is output.

[0024] The method of allocating each route in the drone flight route set to each drone to obtain the drone flight route within a single docking anchor point includes:

[0025] Input the anchor point location, drone flight path set, and drone set;

[0026] For each drone, traverse each flight route. If the length of the flight route is less than the remaining range of the drone, the flight route is assigned to the drone and deleted from the drone flight route set; otherwise, the drone returns to the anchor point to charge;

[0027] The remaining flight routes in the UAV flight route set are arranged in descending order according to flight length, and the UAV with the shortest total length of the arranged routes is selected.

[0028] The method of obtaining the vehicle driving route according to all the stop anchor points includes:

[0029] Create an empty vehicle sequence list;

[0030] The anchor point where the current vehicle is located is added as the starting point to the vehicle sequence list, and the anchor point is deleted from the anchor point set;

[0031] Select an anchor point from the remaining anchor points in the anchor point set that can add the least amount of route length to the current vehicle sequence list, add the anchor point to the end of the vehicle sequence list, and delete it from the anchor point set;

[0032] After multiple iterations until the anchor point set is empty, the obtained vehicle sequence list will generate the vehicle driving route solution.

[0033] The method adjusts the flight path of the drone within a single anchor point by using a taboo search algorithm in response to the real-time change of the drone speed, and outputs a local dynamic optimization solution of the drone and vehicle collaboration model, including:

[0034] In each iteration, a task point is randomly selected from each of the two flight routes for exchange, or a task point is randomly selected from one flight route and added to the other flight route. After completion, the two newly generated flight routes are reordered, and the empty flight routes generated by the insertion operation are deleted. Finally, the routes are reallocated for the drones to generate a new local solution.

[0035] Insert the new local solution into the domain and update the tabu table based on the quality of the solution;

[0036] By calculating the cost of each domain solution, the solution with the minimum cost is selected as the new current solution; if the cost of the new current solution is lower than the previous optimal solution, the optimal solution is updated;

[0037] Repeat the above steps until the maximum number of iterations is reached, and finally output the local dynamic optimization solution.

[0038] The objective function of the local dynamic optimization solution is:

[0039]

[0040] In the formula, Anchor point a q The total execution time of the remaining tasks; The longest total flight time of all drones; q For anchor point; The flight path r i The length of z(u k , r i ) represents the flight path r i Is it assigned to drone u k , if so, then z(u k , r i )=1, otherwise z(u k , r i )=0;x(a q , r i ) represents the anchor point a q Is it included in the flight path? i If so, then x(a q , r i )=1, otherwise x(a q , r i )=0;v uav is the flight speed of the drone.

[0041] A task planning and path optimization device for multi-UAV and vehicle collaboration, which is applied to the above-mentioned method, comprises:

[0042] A UAV and vehicle collaboration model building module is used to build a UAV and vehicle collaboration model, which includes multiple mission points, multiple UAVs and at least one vehicle;

[0043] The global static optimization solution acquisition module is used to select the docking anchor points and the task point set corresponding to each docking anchor point using the K-means clustering algorithm; obtain the UAV flight route set for the task point set within the docking anchor point; assign each route in the UAV flight route set to each UAV to obtain the UAV flight route within a single docking anchor point; obtain the vehicle driving route based on all docking anchor points, thereby obtaining the global static optimization solution of the UAV and vehicle collaboration model;

[0044] The local dynamic optimization solution acquisition module is used to adjust the flight path of the drone within a single anchor point according to the real-time changes in the drone speed through the taboo search algorithm, and output the local dynamic optimization solution of the drone and vehicle collaboration model.

[0045] A mission planning and path optimization device for collaboration between multiple UAVs and vehicles, including a memory and a processor;

[0046] The memory is used to store computer program code and transmit the computer program code to the processor;

[0047] The processor is used to execute the method according to the instructions in the computer program code.

[0048] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method described above is implemented.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] The present invention discloses a task planning and path optimization method and system for the collaboration of multiple drones and vehicles. The method fully considers the advantages of the collaborative work of drones and vehicles, and can perform local dynamic optimization on the flight routes of drones within a single anchor point based on the speed changes of the current drones under constraint conditions and dynamic scenarios, such as the speed changes of the drones. The method also reallocates the tasks of the drones according to the optimization results, ensuring that the drones can efficiently complete data collection tasks under speed changes, thereby improving the flexibility of the method and the efficiency of task response, thereby significantly reducing the response delay of the method to user requests and improving the overall performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a flow chart of a task planning and path optimization method for multi-UAV and vehicle collaboration of the present invention.

[0052] Figure 2It is a schematic diagram of a multi-UAV and vehicle collaborative data acquisition system model mentioned in an embodiment of the present invention.

[0053] Figure 3 It is a schematic diagram of the vehicle driving route and anchor points mentioned in the embodiment of the present invention.

[0054] Figure 4 It is a structural block diagram of a task planning and path optimization device for cooperation between multiple UAVs and vehicles of the present invention.

[0055] Figure 5 It is a structural block diagram of a task planning and path optimization device for cooperation between multiple UAVs and vehicles of the present invention. DETAILED DESCRIPTION

[0056] The present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0057] See also Figure 1 , a task planning and path optimization method for multi-UAV and vehicle collaboration, comprising:

[0058] S1. Establish a UAV and vehicle collaboration model, which includes multiple mission points, multiple UAVs and at least one vehicle, such as Figure 2 As shown; the drone performs data collection tasks, and the vehicle is responsible for transporting the drone and providing communication support.

[0059] The present invention is applicable to data collection tasks in a dynamic environment composed of drones and vehicles, and the path planning and task allocation of drones are determined according to the flight capability of drones and the transportation capability of vehicles. The proposed method aims to solve the technical problem of limited efficiency of drone groups in large-scale regional data collection tasks under the fluctuation of drone speed in dynamic scenes. The method optimizes the collaborative efficiency of drones and vehicles through real-time dynamic adaptation of path planning and task scheduling, and reduces the efficiency loss caused by the limitations of drone endurance, communication, computing power and storage space.

[0060] In the target area There are irregularly scattered task points that need to collect data. Let the initial point set containing all points be P0 = {p1, p2, ..., p n}, all initial task point information is known. For each task point p i (i=1, ..., n), its position on the two-dimensional coordinate axis is p i ∈Ω, i = 1, ..., n. There is an edge connecting any two points, and the set of all edges is E = {(p i , p j )|p i , p j∈P, i≠j}. The distance of any edge is a non-negative number, denoted by Represents the task point p i and p j All the task points and their connected edges form a connected graph G = (P, E). In order to simplify the calculation, the distance between two points in the present invention is calculated by straight-line distance. If it is under the condition of real urban road network, it is necessary to read the road data and calculate the shortest driving distance between two points according to the real road. Let the drone set be U = {u1, u2, ..., u m}. The vehicle-mounted multi-drone starts from starting point A start Departure, after completing all tasks, return to this point. In order to facilitate recording the vehicle route and determining the parking point near the target mission point when the vehicle delivers the drone, the present invention introduces anchor points. In the target area Ω, according to the clustering and partitioning results, after discretization processing, a candidate anchor point is set at a certain distance. The set of all candidate anchor points is A = {a1, a2, ..., a l The vehicle path will select some points from these candidate anchor points as docking anchor points, and the vehicle will wait at the docking anchor points for the drone to perform data collection tasks. Let the vehicle path node set be

[0061] In the multi-UAV-vehicle collaborative data collection problem studied in this invention, first, after receiving the command from the central control console, the vehicle will transport m UAVs in sequence to the parking points near the target mission point according to its planned driving route. After the vehicle transports multiple UAVs to a certain anchor point, the UAVs will take off and go to their respective mission points for data collection. In order to facilitate the recording of vehicle routes and determine the parking points near the target mission points, the concept of anchor points is introduced. Figure 3 As shown in the figure, in the target area Ω, according to the clustering and partitioning results, after discretization, a candidate anchor point is set at a certain distance, and the set of all candidate anchor points is A = {a1, a2, ..., a l In order to improve the working efficiency of the system, it is necessary to shorten the path of the vehicle as much as possible. Therefore, it is necessary to select some suitable points from these candidate anchor points as the docking anchor points. The vehicle will wait at the docking anchor points for the drone to perform the data collection task. Let the vehicle path node set be in i = 1, 2, ..., n. After the anchor point is determined, the path planning problem of the vehicle is to start from the starting point, traverse each path node in the set and finally return to the starting point. The present invention uses the symbol r0 to represent the vehicle's driving route, which has only one. The vehicle's driving route r0 is specifically composed of the starting point A start and the selected anchor point, represented by r0 = {A start, a k1 , a k2 , ..., a kn , A start That is, the vehicle starts from the starting point A. start Departure, passing through anchor point a in order k1 Wait, and finally return to starting point A start . Let any two anchor points a ki and a kj The distance between The vehicle moves from anchor point a ki Drive to a kj The time can be recorded as It can be calculated that:

[0062]

[0063] In a single drone k In a single flight, assume that the set of data collection task points that need to be visited is Q = {q1, q2, ..., q n}, The anchor point of takeoff is recorded as a q , based on this mission point set, a drone u will be generated k For a single flight route, it is denoted as r i ={a q ,q1,q2,...,q n , a q}, where a q ∈A c ,q1,...,q n ∈Q.r i It means the drone is moving from anchor point a q Take off and pass through mission points q1 to q in sequence n , and finally return to anchor point a q The number of drones is defined as M. uav , for all global M uav drones, put their flight paths together to get the drone flight path set Therefore, the global route set in the multi-UAV-vehicle collaboration problem is For drones k On the flight path i In the task node q, whether i Fly to q j , the present invention uses auxiliary decision variables To indicate that, if yes, then otherwise, Accordingly, the flight path r can be calculated as follows: i Length

[0064]

[0065] UAV k The flight times for a single trip according to the flight route are as follows:

[0066]

[0067] Due to the limited battery life of the drone, only a few mission data points can be accessed during a single flight. k At anchor point a q There may be multiple flights at each location. According to the data collection task point of the drone, the central control console will run the algorithm assigned to the drone to calculate the drone path and flight number and send instructions. In addition, the binary variable z(u k , r i ) to represent the route r i Is it assigned to drone u k Specifically, if so, then z(u k , r i )=1; otherwise, z(u k , r i )=0. The variable x(a i , r0) to represent the candidate anchor point a q Whether it is selected as a stop anchor point and included in route r0, the auxiliary decision variable x(a q , r i )=1 to indicate the anchor point a q Is it included in a drone single trip route? i If so, then x(a q , r i )=1; otherwise, x(a q , r i )=0. Then the drone u k At anchor point a q The total route length can be calculated by the following formula:

[0068]

[0069] Assume that the drone u k At anchor point a q There are x flights in total. Since the charging time of the drone is ignored in the model assumptions, the drone u k At anchor point a q The total flight time at is calculated as:

[0070]

[0071] S2. Use the K-means clustering algorithm to select the docking anchor points and the task point set corresponding to each docking anchor point; obtain the UAV flight route set for the task point set within the docking anchor point; assign each route in the UAV flight route set to each UAV to obtain the UAV flight route within a single docking anchor point; obtain the vehicle driving route based on all docking anchor points, thereby obtaining the global static optimization solution of the UAV and vehicle collaboration model.

[0072] Before multiple drones and vehicles set out to perform tasks, the central console will plan and allocate routes for multiple drones and vehicles based on the received initial static tasks and the calculated initial static solution. Generating the initial greedy solution is roughly divided into four steps: anchor point selection and division of anchor point task point set, division of route set within anchor point, drone route allocation, and vehicle route planning. The Initial Greedy Solution Construction Algorithm (IGSCA) is the core step of static task planning for solving the problem of multi-drone-vehicle collaborative patrol. The algorithm first inputs the initial static task point set, the anchor point set to be selected, the anchor point position of the current vehicle, and the drone state set. Its main goal is to generate a feasible initial global solution by reasonably dividing and planning the execution routes of drones and vehicles. Although this solution is not necessarily the optimal solution, it provides a basis for subsequent optimization steps.

[0073] The overall algorithm flow is shown in Algorithm 1. First, call Algorithm 2 to select K centroid anchor points based on K-means clustering, and select the nearest anchor point for each data collection task point, output the task point set corresponding to each anchor point, and thus determine the anchor point position to which the vehicle will go. Secondly, loop through each anchor point, and for the task point set within the anchor point; plan a specific route for the drone to execute within each anchor point, so that the drone starts from the anchor point, visits all the task points assigned to the anchor point, and returns to the anchor point. Call Algorithm 3 to divide and obtain the drone flight route set, and call Algorithm 4 to assign each route in the drone flight route set to each drone without duplication, so as to maximize the drone's flight time and task execution efficiency. Finally, call Algorithm 5 to obtain the vehicle driving route for a drone obtained in the above steps. When the vehicle visits each anchor point, it transports the drone in turn according to the planned path, and finally completes the patrol work of all task points, and finally obtains an initial greedy global solution S.

[0074]

[0075] S3. In response to the real-time changes in the speed of the drone, the flight path of the drone within a single anchor point is adjusted through the taboo search algorithm, and the local dynamic optimization solution of the drone and vehicle collaboration model is output.

[0076] Furthermore, the use of the K-means clustering algorithm to select the docking anchor points and the task point set corresponding to each docking anchor point includes:

[0077] Input the set of task points to be served, the set of anchor points to be selected, and the anchor point location of the current vehicle;

[0078] Randomly select K anchor points from the set of anchor points to be selected as the centroid;

[0079] For each task point to be served, assign it to the cluster set with the nearest centroid;

[0080] Calculate the total distance from all the task points to be served in each cluster set to the centroid, select an anchor point with the shortest distance as the new centroid, and continue to allocate the task points to be served;

[0081] After the initial clustering is completed, check whether the distance between the task point to be served in the cluster set and its centroid exceeds the maximum flight distance limit of the drone. If so, put the task point to be served into the task point set to be selected and reselect an anchor point for it.

[0082] If the anchor point where the vehicle is located is not selected as the centroid anchor point, it will be added to the centroid set, and according to the final centroid set, the task points to be served will be assigned one by one to the cluster set where the nearest centroid is located;

[0083] Add each centroid and its corresponding cluster set to the task point set corresponding to the anchor point.

[0084] Algorithm 2, namely the anchor point selection and anchor point task point set division algorithm based on K-means clustering, aims to cluster the initial static task point set, reasonably select the docking anchor point and divide the task point set, so as to provide preliminary static task planning for the coordinated patrol of UAVs and vehicles. The process of the algorithm is as follows: First, the input of the algorithm includes the task point set to be served P0 = {p1, p2, ..., p n}, the anchor point set A = {A start , a1, a2, ..., a l}, the current vehicle's anchor point position a v The goal of the algorithm is to reasonably divide the task point set into various anchor points and generate an anchor point partition task set E. In the initial step of the algorithm, K anchor points are randomly selected from the anchor point set as the initial centroid. Then, for each task point p i, assign it to the cluster set with the nearest centroid. The algorithm will continue to iterate (the iteration ends when the set number of iterations is reached or the cluster centroid no longer changes), and optimize the division of task points by reselecting new centroids. In each iteration, the total distance from the task points in each cluster set to the centroid is calculated, and an anchor point with the shortest distance is selected as the new centroid to continue to assign task points. After completing the initial clustering, the algorithm will check whether the clustering results meet the distance constraints. If the distance between some task points and their centroid exceeds the maximum flight distance limit, these task points will be placed in the candidate task point set P, and a suitable anchor point will be selected for them. Finally, if the anchor point position a where the vehicle is located v If it is not selected as a centroid anchor point, the algorithm will add it to the centroid set, and according to the final centroid set, the task point set will be assigned to the nearest centroid anchor point one by one. Through this algorithm, preliminary anchor point selection and task point set division results can be provided for subsequent path planning, laying the foundation for the collaborative operation of drones and vehicles.

[0085]

[0086]

[0087] Furthermore, the step of obtaining a flight route set of a drone for a mission point set within a docking anchor point includes:

[0088] Input the anchor point position and the task point set corresponding to the anchor point;

[0089] Set a temporary task point set containing all unvisited task points;

[0090] A task point is randomly selected from the temporary task point set and added to the current route. The distance from the anchor point through the route and back to the anchor point is calculated, and it is determined whether the distance exceeds the maximum flight distance of the drone. If not, the task point is deleted from the temporary task point set and the task point is used as the last node position of the current route. If it exceeds, the task point is deleted from the route and the route is added to the drone flight route set.

[0091] After multiple iterations until the temporary mission point set is empty, the UAV flight route set at the anchor point is output.

[0092] Algorithm 3, the intra-anchor route partitioning algorithm, aims to optimize the partitioning of the mission point set corresponding to each anchor point, so as to plan the best flight route for the drone. The specific process can be summarized as follows: The algorithm first inputs the current anchor point position a iand the task point set E covered by the anchor point. In the initialization phase, the algorithm creates an empty route set routes and sets the maximum number of iterations to max_iter. Then, a temporary task point set Q is set, and the task point E is i The concentrated task points are added to Q. Each time, a task point is randomly selected from the temporary task point set Q and added to the current single route route to determine whether the distance from the anchor point through the route route and back to the anchor point exceeds the maximum flight distance d of the drone. max / 2, once exceeded, the task point is deleted from the route, and the route route is added to the route set routes, and then the route route is set to empty. In each iteration, the algorithm selects the task point closest to the current task point from the task point set and adds it to the current route. If the new task point makes the total flight distance exceed the maximum flight distance of the drone, the task point is deleted and the current route is saved in the route set. Through multiple iterations, the algorithm will try to make the drone visit more task points in each flight while satisfying the drone flight distance constraint. The maximum number of iterations is set to max_iter. After max_iter rounds of loops until the temporary task point set Q is empty, all task points in the anchor point will be assigned to different drone flight routes, thereby outputting the drone flight route set routes at the anchor point. This greedy algorithm optimizes the access order of task points through multiple rounds of iterations to ensure that the generated flight route is as efficient as possible. Among them, location is to record the last node position of the current route route.

[0093]

[0094] Furthermore, the step of allocating each route in the drone flight route set to each drone to obtain the drone flight route within a single docking anchor point includes:

[0095] Input the anchor point location, drone flight path set, and drone set;

[0096] For each drone, traverse each flight route. If the length of the flight route is less than the remaining range of the drone, the flight route is assigned to the drone and deleted from the drone flight route set; otherwise, the drone returns to the anchor point to charge;

[0097] The remaining flight routes in the UAV flight route set are arranged in descending order according to flight length, and the UAV with the shortest total length of the arranged routes is selected.

[0098] After dividing the flight routes of multiple drones within a single anchor point, these routes need to be allocated to each drone. In order to improve the system efficiency of multi-drone-vehicle collaborative patrol, a more balanced allocation is required in the solution, and each drone executes tasks in parallel to minimize the execution time of a single anchor point task.

[0099] Algorithm 4, namely the UAV route allocation algorithm, aims to reasonably allocate the divided flight routes for multiple UAVs to maximize the execution efficiency of the patrol mission. The basic process of the algorithm is as follows: First, the algorithm input includes the anchor point position a i , UAV flight route set routes, UAV set U. For each UAV u k , the algorithm will traverse each flight route r i If the length of the flight path Smaller than the drone u k Remaining flight distance (i.e., the distance from the current drone position through all the mission points in the route and back to the anchor point), then the route r i Assigned to drone u k , and delete the route from the UAV flight route set routes. At this time, the UAV executes the task and returns to the anchor point, and sets the flag bit to True. At this time, the UAV position is updated to the anchor point, and the remaining range is fully charged. If the route length exceeds the remaining range of the UAV, the UAV's status will be updated to need to return to the anchor point to charge first. After each route allocation, the UAV's range status will be reset to a fully charged state, ready to execute the next allocated route. The remaining UAV flight routes in the UAV flight route set are sorted in descending order according to the flight length (the distance from the anchor point through all the task points in the route and then back to the anchor point). Each time the flight route set is traversed, if there is a route that makes the total route length currently arranged for a certain UAV the shortest, then the route is added to the solution set of the UAV to ensure balanced distribution of tasks and reduce the workload of a single UAV. The algorithm can maximize the system efficiency when multiple UAVs perform tasks in collaboration and reduce the total execution time of single anchor point tasks. Finally, the algorithm outputs the task allocation results of each UAV to ensure that the range of each UAV is fully utilized and achieve the global optimal route allocation.

[0100]

[0101]

[0102] Furthermore, obtaining the vehicle driving route according to all the stop anchor points includes:

[0103] Create an empty vehicle sequence list;

[0104] The anchor point where the current vehicle is located is added as the starting point to the vehicle sequence list, and the anchor point is deleted from the anchor point set;

[0105] Select an anchor point from the remaining anchor points in the anchor point set that can add the least amount of route length to the current vehicle sequence list, add the anchor point to the end of the vehicle sequence list, and delete it from the anchor point set;

[0106] After multiple iterations until the anchor point set is empty, the obtained vehicle sequence list will generate the vehicle driving route solution.

[0107] Algorithm 5, the vehicle path routing algorithm, is mainly used to determine the vehicle's driving route between multiple anchor points to minimize the total path length and improve the efficiency of task completion. The basic process of the algorithm is as follows: First, input the initial global solution S, and establish an empty vehicle order list trunk_order, and assign the vehicle's current anchor point A to the vehicle. start It is added to the list as the starting point and deleted from the anchor point set. Then, the algorithm enters the loop traversal step. In each loop, an anchor point is selected from the remaining anchor point set so that its addition to the current path can minimize the increase in route length (the distance from all anchor points in trunk_order to the warehouse). The selected anchor point is added to the end of the trunk_order list and deleted from the anchor point set. This process is repeated until the anchor point set is empty and all anchor points are assigned. The resulting path sequence list generates a solution for the vehicle's route. The vehicle will visit the anchor points one by one in the order of the solution to complete all tasks. The algorithm ensures that the vehicle's route is as short as possible, improving the execution efficiency of the entire system.

[0108]

[0109] Furthermore, in response to the real-time changes in the speed of the drone, the flight path of the drone within a single anchor point is adjusted through the taboo search algorithm, and the local dynamic optimization solution of the drone and vehicle collaboration model is output, including:

[0110] In each iteration, a task point is randomly selected from each of the two flight routes for exchange, or a task point is randomly selected from one flight route and added to the other flight route. After completion, the two newly generated flight routes are reordered, and the empty flight routes generated by the insertion operation are deleted. Finally, the routes are reallocated for the drones to generate a new local solution.

[0111] Insert the new local solution into the domain and update the tabu table based on the quality of the solution;

[0112] By calculating the cost of each domain solution, the solution with the minimum cost is selected as the new current solution; if the cost of the new current solution is lower than the previous optimal solution, the optimal solution is updated;

[0113] Repeat the above steps until the maximum number of iterations is reached, and finally output the local dynamic optimization solution.

[0114] The TS-based Dynamic Processing Optimization Algorithm (TSDPOA) based on taboo search is mainly used for dynamic adjustment and optimization of UAV flight routes. Tabu Search (TS) is an optimization search algorithm based on neighborhood optimization. It records movements through a taboo table to avoid falling into local optimality after multiple iterations. In response to the change in UAV speed, the present invention uses a taboo search algorithm to adjust the UAV route within a single anchor point and output a local optimized solution. The algorithm process is as follows: First, input the solution S within the current single anchor point and initialize the optimal solution and the current solution S. The taboo table List is initially empty. The algorithm optimizes the current solution through multiple rounds of iterations. Each iteration generates multiple solutions in the neighborhood, and selects the optimal solution from them for update. The specific steps include the following parts:

[0115] (1) Neighborhood generation: In each iteration, the algorithm randomly selects a task point from each of the two routes to exchange, or randomly selects a task point from one route and adds it to the other route. After completion, the two newly generated routes are reordered and the empty routes generated by the insertion operation are deleted. Finally, the drone route allocation algorithm is called to reallocate routes for the drones and generate new local solutions.

[0116] (2) Taboo table update: The newly generated solution is inserted into the neighborhood and the taboo table is updated according to the quality of the solution to prevent the algorithm from falling into the local optimum.

[0117] (3) Optimal solution selection: By calculating the cost of each neighborhood solution, the solution with the lowest cost is selected as the new current solution. If the cost of the new solution is lower than the previous optimal solution, the optimal solution is updated.

[0118] (4) Iterative optimization: The algorithm repeats the above steps until the maximum number of iterations is reached, and finally outputs a local optimized solution.

[0119] The algorithm generates multiple candidate solutions and performs taboo search through two basic neighborhood search operations: node exchange and node insertion. After multiple iterations, TSDPOA can effectively optimize the flight path of the UAV in a dynamic environment and ensure the system's mission execution efficiency.

[0120]

[0121]

[0122] In actual situations, due to factors such as terrain and weather, the speeds of drones and vehicles are not constant. Vehicle speed fluctuations have little effect on anchor point division, vehicle routes, and drone routes, but will change vehicle travel time, thereby causing certain fluctuations in the total time cost. When the drone speed fluctuates, it may cause the current single drone planned route to fail to complete or be completed ahead of schedule, thereby causing major changes to the drone route planning and allocation scheme within the current single anchor point. Therefore, the present invention mainly performs dynamic scheme optimization and adjustment for drone speed changes, that is, local optimization and adjustment within the anchor point. Each dynamic processing only requires route optimization within the anchor point where the drone is currently performing a task.

[0123] After every time τ, the current state values ​​of multiple drones will be read, the executed mission points will be deleted, and the flight routes of the drones will be replanned and reallocated for the remaining mission points in the anchor point. Therefore, in order to evaluate the pros and cons of the new plan after replanning and reallocation, it is necessary to calculate the cost of the new plan. q The total execution time of the remaining tasks as optimization goal.

[0124] According to the drone k At anchor point a q The total flight time at the location can be compared to get the longest total flight time of all drones That is to say Then, the optimization objective function is expressed as follows:

[0125]

[0126] In the formula, Anchor point a q The total execution time of the remaining tasks; The longest total flight time of all drones; q For anchor point; The flight path r i The length of z(u k , r i ) represents the flight path r i Is it assigned to drone u k , if so, then z(u k , r i )=1, otherwise z(u k , r i )=0;x(a q , r i ) represents the anchor point a q Is it included in the flight path? iIf so, then x(a q , r i )=1, otherwise x(a q , r i )=0;v uav is the flight speed of the drone.

[0127] In the dynamic scenario where the speeds of multiple drones and vehicles are constantly fluctuating, the complete dynamic task processing algorithm mainly includes: constructing a global initial solution based on K-means clustering, HGA static, and local dynamic optimization based on taboo search (TS). Therefore, the dynamic processing optimization algorithm based on the speed changes of drones and vehicles is called the KGTSA algorithm. According to the principle of starting from the simplest to the most complex, the vehicle-machine joint optimization problem is divided into three steps: problem preprocessing, intra-cluster solution construction, and global solution construction. Among them, the construction of the global initial solution, HGA static optimization, and local dynamic optimization are all optimized and adjusted based on the previous two steps. The overall dynamic simulation process of the KGTSA algorithm is as follows: After the central console receives the initial static task, it calls Algorithm 1-Algorithm 5 to generate a static optimization solution, so that multiple drones and vehicles can start from the starting point to receive the task. After departure, the speed of multiple drones and vehicles has been in random fluctuations, and the random fluctuations are limited to a certain range. After each short period of time τ, if the vehicle is on the way to the next anchor point, the travel distance of the vehicle after the time τ is updated forward; otherwise, it means that the vehicle is parked at the anchor point and the drone is out performing a mission, then the flight distance of each drone after the time τ is traversed and updated forward, and Algorithm 6 is called to generate a local dynamic optimization solution within the current anchor point, and all drones continue to perform tasks according to the new flight route. Repeat the above process until all task points have been visited, and the algorithm ends. Based on the speed change of the current drone, the present invention uses the KGTSA algorithm to perform local dynamic optimization on the flight route of the drone within a single anchor point, and reallocates the tasks of the drone according to the optimization results to ensure that the drone can efficiently complete the data collection task under speed changes.

[0128] The prototype code of the method of the present invention is implemented in Python language, and the feasibility and effectiveness are verified on simulation data.

[0129] See also Figure 4 The present invention also provides a task planning and path optimization device for cooperation between multiple UAVs and vehicles, which is applied to the task planning and path optimization method for cooperation between multiple UAVs and vehicles described above, and the device comprises:

[0130] A UAV and vehicle collaboration model building module is used to build a UAV and vehicle collaboration model, which includes multiple mission points, multiple UAVs and at least one vehicle;

[0131] The global static optimization solution acquisition module is used to select the docking anchor points and the task point set corresponding to each docking anchor point using the K-means clustering algorithm; obtain the UAV flight route set for the task point set within the docking anchor point; assign each route in the UAV flight route set to each UAV to obtain the UAV flight route within a single docking anchor point; obtain the vehicle driving route based on all docking anchor points, thereby obtaining the global static optimization solution of the UAV and vehicle collaboration model;

[0132] The local dynamic optimization solution acquisition module is used to adjust the flight path of the drone within a single anchor point according to the real-time changes in the drone speed through the taboo search algorithm, and output the local dynamic optimization solution of the drone and vehicle collaboration model.

[0133] See also Figure 5 ,The present invention also provides a task planning and path optimization device for cooperation between multiple UAVs and vehicles, including a memory and a processor;

[0134] The memory is used to store computer program code and transmit the computer program code to the processor;

[0135] The processor is used to execute the above-mentioned method for task planning and path optimization of multi-UAV and vehicle collaboration according to the instructions in the computer program code.

[0136] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned method of task planning and path optimization for collaboration between multiple UAVs and vehicles is implemented.

[0137] Generally speaking, the computer instructions for implementing the method of the present invention may be carried in any combination of one or more computer-readable storage media. Non-transitory computer-readable storage media may include any computer-readable media, except for the signal itself that is temporarily propagating.

[0138] The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EKROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, device, or device.

[0139] Computer program code for performing the operation of the present invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages, in particular, Python suitable for neural network computing and platform frameworks based on TensorFlow, PyTorch, etc. can be used. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer or to an external computer (for example, using an Internet service provider to connect via the Internet) through any type of network, including a local area network (LAN) or a wide area network (WAN).

[0140] The above-mentioned device and non-temporary computer-readable storage medium can refer to the specific description of a task planning and path optimization method for multi-UAV and vehicle collaboration and its beneficial effects, which will not be repeated here.

[0141] Although the embodiments of the present invention have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A method for task planning and path optimization for multi-UAV and vehicle collaboration, characterized in that: include: Establish a UAV and vehicle collaboration model, which includes multiple mission points, multiple UAVs and at least one vehicle; The K-means clustering algorithm is used to select the docking anchor points and the task point set corresponding to each docking anchor point; the UAV flight route set is obtained for the task point set within the docking anchor point; each route in the UAV flight route set is assigned to each UAV to obtain the UAV flight route within a single docking anchor point; the vehicle driving route is obtained based on all the docking anchor points, thereby obtaining the global static optimization solution of the UAV and vehicle collaboration model; In response to the real-time changes in the UAV's speed, the UAV's flight path within a single anchor point is adjusted through the taboo search algorithm, and the local dynamic optimization solution of the UAV and vehicle collaboration model is output.

2. The method for task planning and path optimization of multi-UAV and vehicle collaboration according to claim 1, characterized in that: The method of selecting the docking anchor points and the task point set corresponding to each docking anchor point by using the K-means clustering algorithm includes: Input the set of task points to be served, the set of anchor points to be selected, and the anchor point location of the current vehicle; Randomly select K anchor points from the set of anchor points to be selected as the centroid; For each task point to be served, assign it to the cluster set with the nearest centroid; Calculate the total distance from all the task points to be served in each cluster set to the centroid, select an anchor point with the shortest distance as the new centroid, and continue to allocate the task points to be served; After the initial clustering is completed, check whether the distance between the task point to be served in the cluster set and its centroid exceeds the maximum flight distance limit of the drone. If so, put the task point to be served into the task point set to be selected and reselect an anchor point for it. If the anchor point where the vehicle is located is not selected as the centroid anchor point, it will be added to the centroid set, and according to the final centroid set, the task points to be served will be assigned one by one to the cluster set where the nearest centroid is located; Add each centroid and its corresponding cluster set to the task point set corresponding to the anchor point.

3. The method for task planning and path optimization of multi-UAV and vehicle collaboration according to claim 1, characterized in that: The step of obtaining a UAV flight route set for a mission point set within a docking anchor point includes: Input the anchor point position and the task point set corresponding to the anchor point; Set a temporary task point set containing all unvisited task points; A task point is randomly selected from the temporary task point set and added to the current route. The distance from the anchor point through the route and back to the anchor point is calculated, and it is determined whether the distance exceeds the maximum flight distance of the drone. If not, the task point is deleted from the temporary task point set and the task point is used as the last node position of the current route. If it exceeds, the task point is deleted from the route and the route is added to the drone flight route set. After multiple iterations until the temporary mission point set is empty, the flight route set of the UAV at the anchor point is output.

4. The method for task planning and path optimization of multi-UAV and vehicle collaboration according to claim 1, characterized in that: The method of allocating each route in the drone flight route set to each drone to obtain the drone flight route within a single docking anchor point includes: Input the anchor point location, drone flight path set, and drone set; For each drone, traverse each flight route. If the length of the flight route is less than the remaining range of the drone, the flight route is assigned to the drone and deleted from the drone flight route set; otherwise, the drone returns to the anchor point to charge; The remaining flight routes in the UAV flight route set are arranged in descending order according to flight length, and the UAV with the shortest total length of the arranged routes is selected.

5. The method for task planning and path optimization of multi-UAV and vehicle collaboration according to claim 1, characterized in that: The method of obtaining the vehicle driving route according to all the stop anchor points includes: Create an empty vehicle sequence list; The anchor point where the current vehicle is located is added as the starting point to the vehicle sequence list, and the anchor point is deleted from the anchor point set; Select an anchor point from the remaining anchor points in the anchor point set that can add the least amount of route length to the current vehicle sequence list, add the anchor point to the end of the vehicle sequence list, and delete it from the anchor point set; After multiple iterations until the anchor point set is empty, the obtained vehicle sequence list will generate the vehicle driving route solution.

6. The method for task planning and path optimization of multi-UAV and vehicle collaboration according to claim 1, characterized in that: The method adjusts the flight path of the drone within a single anchor point by using a taboo search algorithm in response to the real-time change of the drone speed, and outputs a local dynamic optimization solution of the drone and vehicle collaboration model, including: In each iteration, a task point is randomly selected from each of the two flight routes for exchange, or a task point is randomly selected from one flight route and added to the other flight route. After completion, the two newly generated flight routes are reordered, and the empty flight routes generated by the insertion operation are deleted. Finally, the routes are reallocated for the drones to generate a new local solution. Insert the new local solution into the domain and update the tabu table based on the quality of the solution; By calculating the cost of each domain solution, the solution with the minimum cost is selected as the new current solution; if the cost of the new current solution is lower than the previous optimal solution, the optimal solution is updated; Repeat the above steps until the maximum number of iterations is reached, and finally output the local dynamic optimization solution.

7. The method for task planning and path optimization of multi-UAV and vehicle collaboration according to claim 6, characterized in that: The objective function of the local dynamic optimization solution is: In the formula, Anchor point a q The total execution time of the remaining tasks; The longest total flight time of all drones; q For anchor point; The flight path r i The length of z(u k , r i ) represents the flight path r i Is it assigned to drone u k , if so, then z(u k , r i )=1, otherwise z(u k , r i )=0;x(a q , r i ) represents the anchor point a q Is it included in the flight path? i If so, then x(a q , r i )=1, otherwise x(a q , r i )=0;v uav is the flight speed of the drone.

8. A task planning and path optimization device for multi-UAV and vehicle collaboration, characterized in that: The device is applied to the method described in any one of claims 1 to 7, and the device comprises: A UAV and vehicle collaboration model building module is used to build a UAV and vehicle collaboration model, which includes multiple mission points, multiple UAVs and at least one vehicle; The global static optimization solution acquisition module is used to select the docking anchor points and the task point set corresponding to each docking anchor point using the K-means clustering algorithm; obtain the UAV flight route set for the task point set within the docking anchor point; assign each route in the UAV flight route set to each UAV to obtain the UAV flight route within a single docking anchor point; obtain the vehicle driving route based on all docking anchor points, thereby obtaining the global static optimization solution of the UAV and vehicle collaboration model; The local dynamic optimization solution acquisition module is used to adjust the flight path of the drone within a single anchor point according to the real-time changes in the drone speed through the taboo search algorithm, and output the local dynamic optimization solution of the drone and vehicle collaboration model.

9. A task planning and path optimization device for multi-UAV and vehicle collaboration, characterized in that: including memory and processor; The memory is used to store computer program code and transmit the computer program code to the processor; The processor is configured to execute the method according to any one of claims 1 to 7 according to instructions in the computer program code.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images