Path planning method for heterogeneous cooperation system of unmanned aerial vehicle and unmanned vehicle

Abstract

The application provides a path planning method of an unmanned aerial vehicle and unmanned vehicle heterogeneous cooperative system, which comprises a step of generating an unmanned aerial vehicle dynamic transfer path, in which the unmanned aerial vehicle transfers from a current gathering point to a next gathering point through at least one task point. The technical solution of the application adopts an unmanned aerial vehicle dynamic gathering mode, that is, in the unmanned aerial vehicle and unmanned vehicle heterogeneous cooperative system, when the system needs to transfer to a next task point, the unmanned vehicle and the unmanned aerial vehicle take different transfer paths at the same time, the unmanned vehicle passes through a task point of the unmanned vehicle, while the unmanned aerial vehicle passes through a task point of the unmanned aerial vehicle at the same time, after performing a task point access task, the unmanned aerial vehicle goes to the next task point to meet the unmanned vehicle. In the dynamic gathering process, the transfer step is changed from only performing a transfer process to synchronously completing a task point transfer in the process of accessing the task point, so that the working time of the unmanned aerial vehicle can be compressed, the working time of the unmanned aerial vehicle and unmanned vehicle heterogeneous cooperative system is shortened, and the system operation efficiency is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of path planning methods for unmanned systems, in particular to a path planning method for a heterogeneous cooperative system of unmanned aerial vehicles and unmanned ground vehicles. BACKGROUND

[0002] Due to the good complementarity between unmanned aerial vehicles (UAVs) and unmanned ground vehicles (UGVs), it has good practical significance to study a UAV-UGV cooperative working system, which can be widely used in military, disaster relief, detection and other fields. In such a UAV-UGV cooperative working system, the UGV is limited by its own walking mechanism and cannot reach all task points; and the UAV is also limited by the energy of its own battery and cannot reach all task points with one charge. Therefore, the UGV needs to serve as an energy supply station for the UAV for take-off and landing, which is also a significant feature of such a UAV-UGV cooperative working system.

[0003] Ropero published a paper entitled "TERRA: A path planning algorithm for cooperative UGV-UAV exploration" in 2018, which proposed a path planning algorithm for UGV-UAV cooperative work. This method converts the problem of moving the UGV between charging points into a traveling salesman problem to obtain the driving path of the UGV, and converts the problem of traversing the UAV between charging points and task points into a TSPs model and uses the A* pathfinding algorithm to solve the motion path of the UAV. In the model established by Ropero, the UGV is only used as a charging station for the UAV, and the movement process between the UGV and the UAV is carried out separately. When the UGV performs the task of traversing the task points, the UGV is stationary at the current position and waits for the UAV to return before going to the next charging point. During the process of going to the next task point, the UAV is in a resident state on the UGV. Since the UAV is in a resident state during the process of going to the next task point, it does not participate in work during the resident period. Therefore, although this method ensures the optimal route to a certain extent, it is not optimal in terms of actual system running time due to the existence of the resident time. SUMMARY

[0004] In view of the problem that the prior art solution is not optimal in terms of actual system running time, the present application provides a path planning method for a heterogeneous cooperative system of unmanned aerial vehicles and unmanned ground vehicles with parallel characteristics.

[0005] The technical solution of the present application is a path planning method for a heterogeneous cooperative system of unmanned aerial vehicles and unmanned ground vehicles, which includes a step of generating a dynamic transfer path for the unmanned aerial vehicles. In the dynamic transfer path for the unmanned aerial vehicles, the unmanned aerial vehicles go from the current rendezvous point to the next rendezvous point through at least one task point.

[0006] Specifically, the step of generating the dynamic transfer path of the UAV adopts a hybrid genetic algorithm to determine an optimal solution population with a minimum path length as an optimization objective, and the optimal solution population is a set of UAV paths in a current genetic iteration process.

[0007] Specifically, in the genetic iteration process of the hybrid genetic algorithm, an infeasible solution is introduced, and the optimal solution population is updated by an adaptive large neighborhood search algorithm.

[0008] Specifically, in the hybrid genetic algorithm process, the UAV advancing state of each task point position in the UAV path is encoded respectively, and the set of the advancing state includes a UAV advancing to a next task point state, a UAV returning to an original rendezvous point state, and a UAV advancing to a next rendezvous point state.

[0009] Specifically, in the hybrid genetic algorithm process, the cross variation operation on the UAV path is further included, and the cross variation operation includes:

[0010] all the UAV paths in the optimal solution population are divided into three sub-path segments, which are a transfer path segment, a return first rendezvous point segment, and a return second rendezvous point segment;

[0011] the path segments are divided into path groups, all the transfer path segments are divided into a transfer path group, all the return first rendezvous point segments are divided into a return first rendezvous point group, and all the return second rendezvous point segments are divided into a return second rendezvous point group;

[0012] the cross variation operation between groups and / or the cross variation operation within groups is performed.

[0013] Specifically, the cross variation operation between groups is a random exchange operation of genetic operators on the sub-path segments belonging to different path groups, and the cross variation operation within groups is a random exchange operation of genetic operators on the sub-path segments belonging to the same path group. The genetic operator is the task point in the sub-path segment.

[0014] Specifically, the A* algorithm is adopted to determine an initial population of the hybrid genetic algorithm.

[0015] Preferably, the path planning method of the heterogeneous collaborative system of the UAV and the unmanned vehicle further includes a step of generating a rendezvous point with an energy limit of the UAV as a constraint condition and a shortest path of the unmanned vehicle as an optimization objective.

[0016] Specifically, the step of generating the rendezvous point includes a step of determining a rendezvous point candidate set by a V-Map division method.

[0017] Specifically, the generating set-up node step further comprises a step of determining a set-up node preferred set in the set-up node candidate set by a minimum set cover method.

[0018] Specifically, the generating set-up node step further comprises a step of determining a set-up node set in the set-up node preferred set by a gradient vector optimization method.

[0019] Specifically, the path planning method of the UAV-UGV heterogeneous collaborative system further comprises a step of determining a UGV working path of the UGV traversing the UGV task points and the set-up nodes by an LKH algorithm with the shortest path of the UGV as an optimization target.

[0020] Preferably, the path planning method of the UAV-UGV heterogeneous collaborative system further comprises a step of determining a UAV working path of the UAV traversing the UAV task points in the sub-task area by an A* algorithm with the shortest combined path of the UAV in the sub-task area as an optimization target.

[0021] The technical scheme of the present application adopts a parallel working mode of dynamic set-up of UAVs, that is, in the UAV-UGV heterogeneous collaborative system, when the system needs to be transferred to a next task point, the UGV and the UAV take different transfer paths at the same time, the UGV is transferred to a next set-up node via a UGV task point, and the UAV is transferred to the next set-up node via a UAV task point after performing a task point access task. In the dynamic set-up process, the transfer step is changed from only performing a transfer process to synchronously completing a task point transfer in a task point access process, so that the transfer time of the UAV-UGV heterogeneous collaborative system can be compressed, the actual working time of the system can be optimized, and the system operation efficiency can be improved. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 The figure is a general flowchart of the present application;

[0023] Figure 2 The figure is a flowchart of the generating set-up node process of the present application;

[0024] Figure 3 The figure is a working timing diagram of the UAV-UGV heterogeneous collaborative model of the present application.

[0025] 1: generating set-up node 2: determining UGV working path 3: determining UAV working path 11: generating set-up node candidate set S cor,tmp 12: generating set-up node preferred set S cor,car 13: generating set-up node set S cor31: A* algorithm generates initial population 32: Calculate fitness 33: Group individuals 34: Cross mutation to generate next generation population 35: ALNS algorithm optimizes next generation population 36: Determine whether the maximum number of iterations is reached DETAILED DESCRIPTION

[0026] The application will be described in detail below in combination with the drawings and specific embodiments. In the present specification, the size proportions of the drawings do not represent actual size proportions, and they are only used to reflect the relative position relationship and connection relationship between the components. Components with the same name or the same label represent similar or identical structures, and are only for illustrative purposes.

[0027] UAV is Unmanned Aerial Vehicle, and UGV is Unmanned Ground Vehicle. Therefore, the UAV-UGV heterogeneous collaborative system is referred to as UAV-UGV collaborative system hereinafter. It is believed that in the UAV-UGV collaborative system, the total time requirement is often more reasonable than the total distance requirement. On the one hand, in actual conditions, the total time requirement is often a strong constraint condition specified in advance in the task demand, and the total distance constraint is a target refinement design in the task design to achieve the total time requirement and other constraint conditions; on the other hand, for the model mentioned in the background art that takes the total distance as the optimization target, due to the existence of non-task time such as residence time, the total time is often much greater than the theoretical working time calculated according to the total path. Specifically, in the working model of the background art, when the UGV carries the UAV to the next rendezvous point, the UAV cannot perform the motion of visiting the task point; the residence time of this part does not bring any beneficial effect to the task of the UAV visiting the task point. Therefore, even in the case of the shortest total path, the system running time is actually the simple sum of the running time of the UAV and the running time of the UGV, and if the two can move simultaneously, the total time has a lot of room for improvement.

[0028] Based on such consideration, the present application first improves the existing technical model and proposes a new working model of the UAV-UGV collaborative system with parallel characteristics. The input of the working model of the UAV-UGV collaborative system is:

[0029] 1. Only the case of single UAV and single UGV collaboration is considered;

[0030] 2. A given number m of UAV task point set P uav , where the i-th task point is denoted as p uav,i ;

[0031] 3. A given number n of UGV task point set P ugv , where the j-th task point is denoted as pugv,j ;

[0032] 4. Consider that the running speed of UAV and UGV is constant respectively, and the movement speed of UAV is much greater than that of UGV;

[0033] 5. Only consider that the running of UAV has energy constraint, which actually limits the longest distance d of single running of UAV max ;

[0034] 6. UAV and UGV can move simultaneously, and the movement path can be coordinated.

[0035] Solving the shortest working time problem of the system under the condition of the model can actually be divided into two sub-problems. One is to consider that the speed of UAV is much greater than the movement speed of UGV, so it is necessary to find a minimum number of rendezvous point sets that can guarantee to complete the access to all task points under the condition of assigning the movement path to UAV as much as possible, and to generate the shortest working path of UGV. On the other hand, for the above determined rendezvous point set, it is necessary to determine the working path of UAV including the dynamic transfer path of UAV. That is, to generate the working path of the system that meets the shortest running time goal of the UAV-UGV cooperative system under the above constraints, which not only contains the working path of UGV, but also contains the working path of UAV.

[0036] In the above working model of UAV-UGV cooperative system, the pre-task of generating the working path is to generate the rendezvous point set S cor , and the number of rendezvous points is l, and the kth rendezvous point in it can be denoted as s cor,k In some tasks, the rendezvous point set can be given in advance, so that the process of generating the rendezvous point can be omitted. Of course, more often, the system running time can be shortened and the system running efficiency can be improved by regenerating the rendezvous point set, and obviously there is also a situation without the given rendezvous point set, so that only regeneration is allowed.

[0037] After the rendezvous point set S cor is determined, the working state of the model of the UAV-UGV cooperative system according to the technical scheme of the present application can actually be divided into two types:

[0038] The first is the parallel access state, also known as the dynamic transfer process, that is, the process of dynamic transfer of the UAV-UGV cooperative system from the current rendezvous point to the next rendezvous point. At this time, the UGV will start from the current rendezvous point, pass through the task points it plans to visit, and arrive at the next rendezvous point, or in some models and actual situations, the UGV will not be set with task points, and the UGV will start from the current task point and directly go to the next rendezvous point. Unlike the prior art, in which the UAV resides on the UGV and arrives at the next rendezvous point with the UGV, in the parallel access state, the UAV starts from the current rendezvous point and synchronously goes to the next task point after passing through part of the task points of the UAV under the condition of meeting the energy constraint. The transfer of the UAV and the UGV is synchronized, and the transfer time is actually determined by the maximum value of the transfer time T uav,para of the UAV and the transfer time T ugv,para of the UGV. Of course, it needs to be explained that we say that in the process of parallel access, the UAV will complete the access of part of the task points, which means that after calculation and evaluation, it is determined according to the calculation result which task points will be accessed in the transfer process. If the calculation and evaluation result is that the UAV cannot achieve synchronous access to the task points or even parallel access to the task points does not have a time advantage, and further makes the system make the UAV make a direct action from the current rendezvous point to the next rendezvous point in the current transfer process, it should also be regarded as the dynamic transfer step generated in this application has been executed.

[0039] The other is the serial access state, that is, the UGV is located at the current rendezvous point and waits, and the UAV is responsible for completing the access to the task points corresponding to the current rendezvous point that are not in the parallel access process with the current rendezvous point as the base point. In the process of executing the access task, the UAV may need to go back and forth between the current rendezvous point and the task point multiple times, and the UGV always waits for the UAV at the current rendezvous point. The working path of the UAV in this serial access state needs to be considered in three independent parts, the first is the task points in the part overlapping with the task area of the previous rendezvous point, the second is the task points in the part overlapping with the task area of the next rendezvous point, and the last is the task points in the part covered by the task area of the current rendezvous point alone. The task area here can be regarded as a circular area with the rendezvous point as the center and the maximum circular area that the UAV can go back and forth. The first and second parts above are actually two aspects of the same problem, that is, they are overlapping task areas, which can be solved together when generating the dynamic transfer path of the UAV. We will explain in detail the solution to the two conditions corresponding to the above three parts later.

[0040] In the model of the present application, since it has been assumed that the respective operating speeds of the UAV and the UGV are constant, the operating time of a certain operating time period is proportional to the motion distance during the period, and therefore the local path distance can be used instead of the local operating time. Therefore, the optimization objective of the shortest time can still be converted into the shortest path problem in the working model of the UAV and UGV cooperative system of the present application, and the final system working time actually depends on the cooperation between the UAV and the UGV, i.e., when moving synchronously, the longer time of the two is taken as the criterion.

[0041] In summary, referring to the flowchart of Figure 1 , the overall solution process of the technical solution of the present application is to first generate the set of set-up points 1, and then complete the steps of determining the UGV operating path 2 and determining the UAV operating path 3 according to the set of set-up points that has been determined.

[0042] Set-up point set S cor that meets the model requirements k , in addition to achieving the optimization objective of the shortest operating path of the UGV, it is also necessary to ensure that the round trip of the UAV at all set-up points can ensure access to all task points. For a given set-up point s max , the task points within the circular region surrounded by the maximum access radius R max of the UAV are all accessible. The region corresponding to the set-up point s k can be recorded as the sub-task region P k at the set-up point, which is actually a subset of the total task region; the set of UAV task points in the region is recorded as P uav (s k ), then for any UAV task point p uav,i , as long as p uav,i ∈P uav, (s k ), the UAV can access p k with s uav,i as the base point. Based on this idea, in the above step of generating set-up points 1, the set-up point set S cor that meets the conditions and makes the UGV path shortest is framed step by step through the following steps.

[0043] 11 Generate set-up point candidate set S cor,tmp . The purpose of this step is to generate a set-up point candidate set S cor,tmp containing the set-up point set, and the specific implementation is to obtain a Voronoi diagram corresponding to all task points of the UAV by the Voronoi diagram division method, and all endpoints of the Voronoi diagram are the set-up point candidate set S cor,tmp .

[0044] 12 Generate set-up point preferred set S cor,carThe step finds the minimum number of R max One of the covering methods is to cover all the task points of the UAV by circular areas with radius, so for each task point in the circular area, there is always a point that can be used as the task base point of the UAV in the area, so the above-mentioned candidate set S cor,tmp of the rendezvous point in the area can be selected to form a new preferred set S cor,car .

[0045] 13 Generate the rendezvous point set S cor Since it is assumed that the speed of the UGV is much smaller than that of the UAV, the time of the UGV has a more significant impact on the running time of the system, so in order to shorten the running time of the system, the distance of the UGV should be shortened as much as possible and the distance of the UAV should be extended as much as possible. In other words, as long as the range of access can be optimized by the UAV path, do not access in the way of increasing the distance between the rendezvous points. Therefore, this step optimizes the preferred set S cor,car of rendezvous points by introducing a gradient vector based on the preferred set of rendezvous points, so that the distance between the rendezvous points is as far as possible under the constraint condition, and the final result is used as the determined rendezvous point set S cor .

[0046] In the above-mentioned step of generating the rendezvous point 1, the determined rendezvous point set S cor is finally obtained by step-by-step optimization. Through step-by-step optimization, for any one rendezvous point s k , it can always be used as a base point to realize access to the UAV task points P k (s uav, ) in its corresponding sub-task area P k .

[0047] The following process calculates the shortest paths for both the UGV and UAV, based on known node points. As discussed later, some steps in the UAV path generation process are related to the order of node points traversed in the UGV path; therefore, the calculation of determining the UGV working path 2 needs to be completed before determining the UAV working path 3. The problem of finding the shortest path for the UGV given known node points—that is, finding the minimum path without return for a given set of points—is essentially a Traveling Salesman Problem (TSP). The background paper has already provided an algorithmic implementation using a fast route-first heuristic to solve this problem and obtain the minimum UGV path. Furthermore, existing technologies for solving the TSP include backtracking, dynamic programming, genetic algorithms, and neural network algorithms, all of which have proven effective and can be applied to the UGV path planning process in this application. The following focuses on the UAV path planning problem in this application.

[0048] In this application, since the model already specifies that UAVs and UGVs can move simultaneously and their movement paths can be coordinated, one of the key points is how to determine the coordinated path during simultaneous movement. In the following description, we use a dynamic assembly process to represent the process of UGVs and UAVs moving simultaneously. This is natural because during the dynamic assembly process, the events that occur simultaneously are the UGV moving from the current assembly point to the next assembly point via the UGV dynamic transfer path and the UAV moving from the current assembly point to the next assembly point via the UAV dynamic transfer path.

[0049] In generating node 1, we propose a subtask region P through the process of generating node 1. uav (s k The concept of ) and the corresponding node s k The base point is the location of the task area. Within this sub-task area, one or more UAV movement paths can always be found that travel between the task point and the base point; this is guaranteed in advance through the node generation process. After determining the UGV work path 2, and the UGV's movement path, the corresponding order of the node points traversed by the UGV and UAV during their movement is also determined. For two adjacent node points s in this node point order... k With s k+1 Their corresponding sub-task regions P uav (s k ) and P uav (s k+1 There may or may not be any overlap between them.

[0050] Consider the case where there is an intersection, denoted by Ω. For a node s... k In the non-intersecting part P of its subtask regionuav (s k ) in the intersection Ω can only be accessed in the serial access state; for the rendezvous point s k+1 , the same conclusion applies; and for the task points in the intersection Ω, there are three cases: one is accessed with s k as the base point; two is accessed with s k+1 as the base point; and three is accessed in the UAV dynamic transfer path from the rendezvous point s k to s k+1 . The first and second of these three cases correspond to the task points in the first and second parts of the task area overlap, and the third case is the problem to be solved by the dynamic transfer path. For the task points in the part of the sub-task area outside the intersection Ω, which is equivalent to the third part of the task area overlap, the corresponding case is actually equivalent to the process of traversing the task points of the UAV in the sub-task area in the prior art.

[0051] Considering the case where there is no intersection in the sub-task area, since each sub-task area is independent, it means that the distance between the rendezvous points does not meet the energy constraint condition of the UAV, so the UAV can only reach the next rendezvous point by residing on the UGV, which is no different from the prior art and will not be repeated here. In the prior art, the process of traversing the task points of the UAV in the sub-task area. Corresponding to a class of Traveling Salesman Problem (TSPs), in order to meet the energy constraint condition of the UAV, the UAV may repeatedly return to the base point when accessing the task points in the non-intersection part, forming multiple TSP paths. For this, the present application uses the A* algorithm for processing, and of course other methods for solving TSPs in the prior art are also feasible.

[0052] For the problem of accessing the task points of the UAV in the intersection part Ω of the sub-task area, according to the above analysis, the situation is much more complex, as described earlier, in addition to considering the generation of the dynamic transfer path, the path for accessing the task points in this area with s k as the base point and the path for accessing the task points in this area with s k+1The path to the task points within the intersection Ω is determined by the base point. The A* algorithm is no longer sufficient to meet the computational needs, accuracy requirements, and time efficiency requirements. To generate dynamic transfer paths for the UAV in the Ω region, this application designs a task planning algorithm based on a hybrid genetic algorithm (GA) and an adaptive large neighborhood search algorithm (ALNS). In this algorithm, the GA algorithm performs genetic optimization on the initial solution within a given number of iterations. During each iteration, the ALNS algorithm is used to disrupt and repair infeasible and optimal solution candidate sets. Introducing infeasible solutions expands the search range for optimal solutions, avoiding getting trapped in local optima. Considering the three possible paths for task points in the intersection Ω, the algorithm encodes the forward state of each task point in the path using a code table c∈C, C=[0,1,2], where c=0 indicates the UAV is moving to the next task point, c=1 indicates the UAV is returning to the original node, and c=2 indicates the UAV is moving to the next node. Where c = 1 represents the forward state of the UAV returning to the base point, which actually includes the base point being s. k With s k+1 There is no need to make additional distinction between the two cases in the algorithm. For any task point in the intersection Ω, if it can be accessed by the UAV, then its forward state must belong to one of the three in the code table mentioned above.

[0053] The specific process can be used Figure 3 The flowchart shown is an example.

[0054] Step 31 of the A* algorithm for generating the initial population. In the intersection Ω region of the aforementioned adjacent subtask regions, the A* algorithm is used to perform local optimization to generate the initial path r. i As the initial population, in this step, we only need to consider whether the addition of task points satisfies the drone's endurance constraint, calculate the path length to the next node as the estimated cost, search for the optimal dynamic transfer path segment, and accordingly assign the remaining task points to the path segments returning to the first node and returning to the second node, thereby effectively reducing invalid solutions and pruning the search space, and accelerating the algorithm's convergence process.

[0055] The fitness calculation involves 32 steps. In this step, each individual generated in each iteration is a candidate for the optimal solution. All individuals constitute the optimal solution population for the current iteration. If the current iteration number is denoted as iter, then the population R of the current generation... iter Individual r in iter It should contain three sub-path segments r iter,1 r iter,2 r iter,3 r iter,1 This represents the transition path segment in the candidate, r. iter,2 This indicates that the candidate with "s" in it is...k The sub-path segment (multiple round trips are recorded as a sub-path) taking the base point as the base point is denoted as the first return rendezvous point path segment, r iter,3 The sub-path (in the same way, multiple round trips are recorded as a sub-path) taking the base point as the base point in the candidate is denoted as the second return rendezvous point path segment, r k+1 The sub-path (in the same way, multiple round trips are recorded as a sub-path) taking the base point as the base point in the candidate is denoted as the second return rendezvous point path segment, r iter,1 The length of r iter,2 The length of r iter,3 The length of r iter,1 The length of r iter,2 The length of r iter,3 The length of r

[0056] Individual grouping 33 step. In this step, the sub-path segments of all individuals in the population of this iteration are grouped, that is, all the transfer path segments r iter,1 are a group, the first return rendezvous point path segments r iter,2 are a group, the second return rendezvous point path segments r iter,3 are a group. The operation of this step is mainly to facilitate the subsequent intra-group and / or inter-group crossover and mutation operations. This part is not necessarily formal, and in fact can be integrated into the crossover and mutation operation. Even so, it should be regarded as a grouping operation process implicit in the crossover and mutation process.

[0057] Crossover and mutation to generate the next generation population 34 step. This step is the core step of the GA genetic algorithm, which completes the crossover and mutation operation between individuals. In this step, the operation operator selected by the present application is the task point in each path segment, and the optimal solution population of this genetic iteration is generated by randomly exchanging the task points between the sub-path segments in the sub-path groups that have been grouped in the front. The exchange operation can be selected to be performed within the group and / or between the groups.

[0058] ALNS algorithm optimization of the next generation population 35 step. In the crossover and mutation to generate the next generation population 34 step, since the energy constraint condition of the UAV cannot be considered in the genetic process, there are a large number of infeasible solutions. In a simplified embodiment, the infeasible solutions can be directly deleted and not recorded in the genetic process. This step provides a more optimized implementation, that is, the infeasible solutions are still retained, and the ALNS (adaptive large neighborhood search) algorithm in this step performs destruction and repair operations on the individuals representing the infeasible solutions, so as to realize the expansion of the neighborhood range, and optimizes the new optimal solution set obtained to obtain the best optimal solution individual, which is used as the input of the initial population of the next round of genetic iteration. By ALNS, it is avoided to fall into the local optimal condition, the diversity of the new population is ensured, and finally a good optimal solution population can be generated.

[0059] Step 36, judge whether the maximum iteration number is reached. Confirm whether the genetic iteration number reaches the preset maximum iteration number, if the preset maximum iteration number is reached, exit the genetic algorithm process, and output the optimal solution individual with the best performance in this iteration as the optimal solution. The sub-path segment in the optimal solution represents the transfer path segment corresponding to the unmanned aerial vehicle in the intersection Ω part of the sub-character region, that is, the aforementioned dynamic transfer path, the path segment returning to the first rendezvous point and the path segment returning to the second rendezvous point.

[0060] The above merely describes the preferred embodiments of the present application, and does not limit the scope of the present application. Without departing from the design spirit of the present application, various modifications and improvements to the technical solutions of the present application made by those skilled in the art shall fall within the protection scope determined by the claims of the present application.

Claims

1. A path planning method for a heterogeneous collaborative system of unmanned aerial vehicles and unmanned vehicles, characterized in that, The process includes generating a dynamic transfer path for the UAV, in which the UAV moves from the current assembly point to the next assembly point via at least one task point. The aggregation point is obtained in the following way: Generate a candidate set of rally points: The Veno graph corresponding to all task points of the UAV is obtained by partitioning the Veno graph. All endpoints of the Veno graph are the candidate set of rally points. Generate a preferred set of node points: This step uses the minimum set method to find a way to cover all the task points of the UAV with the fewest number of circular regions with the maximum access radius. Thus, for each task point in the circular region, there is always one point that can be used as the task base point of the UAV in that region. Therefore, points can be selected from the candidate set of node points in that region to form a new preferred set of node points. Generate a set of nodes, optimize the node selection set using gradient vectors, and use the final result as the final set of nodes.

2. The path planning method for a heterogeneous collaborative system of unmanned aerial vehicles and unmanned vehicles as described in claim 1, characterized in that, The step of generating the dynamic transfer path of the UAV uses a hybrid genetic algorithm to determine the optimal solution population with the goal of minimizing the path length. The optimal solution population is the set of UAV paths in the current genetic iteration process.

3. The path planning method for a heterogeneous collaborative system of unmanned aerial vehicles and unmanned vehicles as described in claim 2, characterized in that, During the genetic iteration process of the hybrid genetic algorithm, infeasible solutions are introduced and the optimal solution population is updated using an adaptive large neighborhood search algorithm.

4. The path planning method for a heterogeneous collaborative system of unmanned aerial vehicles and unmanned vehicles as described in claim 2, characterized in that, In the process of the hybrid genetic algorithm, the forward state of the UAV at each task point in the UAV path is encoded separately. The forward state set consists of: UAV moving to the next task point, UAV returning to the original node, and UAV moving to the next node.

5. The path planning method for a heterogeneous collaborative system of unmanned aerial vehicles and unmanned vehicles as described in claim 2, characterized in that, The hybrid genetic algorithm process also includes a crossover and mutation operation on the UAV path, which includes: The paths of all UAVs in the optimal solution population are divided into three sub-path segments: a transfer path segment, a return to the first node segment, and a return to the second node segment. Divide the path groups according to the sub-path segments, divide all the transfer path segments into transfer path groups, divide all the return to the first node group into return to the first node group, and divide all the return to the second node group into return to the second node group. Perform crossover and / or crossover operations between groups and within groups.

6. The path planning method for a heterogeneous collaborative system of unmanned aerial vehicles and unmanned vehicles as described in claim 5, characterized in that, The crossover and mutation operation between groups is a random interchange operation of genetic operators performed on sub-path segments belonging to different path groups; the crossover and mutation operation within a group is a random interchange operation of genetic operators performed on sub-path segments belonging to the same path group, wherein the genetic operator is the task point in the sub-path segment.

7. The path planning method for a heterogeneous collaborative system of unmanned aerial vehicles and unmanned vehicles as described in claim 5, characterized in that, The A* algorithm is used to determine the initial population for the hybrid genetic algorithm.

8. The path planning method for a heterogeneous collaborative system of unmanned aerial vehicles and unmanned vehicles as described in claim 1, characterized in that, The path planning method for the heterogeneous collaborative system of UAVs and unmanned vehicles also includes a step of generating a cluster point with the energy limitation of the UAV as a constraint and the shortest path of the unmanned vehicle as the optimization objective.

9. The path planning method for a heterogeneous collaborative system of unmanned aerial vehicles and unmanned vehicles as described in claim 8, characterized in that, The step of generating cluster points includes determining a candidate set of cluster points using the Venn diagram partitioning method.

10. The path planning method for a heterogeneous collaborative system of unmanned aerial vehicles and unmanned vehicles as described in claim 9, characterized in that, The step of generating cluster points also includes the step of determining the preferred set of cluster points from the candidate cluster points using the minimum set coverage method.

11. The path planning method for a heterogeneous collaborative system of unmanned aerial vehicles and unmanned vehicles as described in claim 10, characterized in that, The step of generating cluster points also includes the step of determining the set of cluster points in the preferred set of cluster points using a gradient vector optimization method.

12. The path planning method for a heterogeneous collaborative system of unmanned aerial vehicles and unmanned vehicles as described in claim 8, characterized in that, The path planning method for the heterogeneous collaborative system of UAVs and unmanned vehicles also includes the step of determining the working path of the unmanned vehicle by using the LKH algorithm, with the shortest path of the unmanned vehicle as the optimization objective.

13. The path planning method for a heterogeneous collaborative system of unmanned aerial vehicles and unmanned vehicles as described in claim 8, characterized in that, The path planning method for the heterogeneous collaborative system of UAVs and unmanned vehicles also includes the step of determining the working path of the UAV traversing the UAV task points within the sub-task area by using the A* algorithm with the shortest combined path of the UAV within the sub-task area as the optimization objective.

Images