Multi-task scheduling method for multiple unmanned platforms

Abstract

The invention relates to a multi-task scheduling method for multiple unmanned platforms, belongs to the technical field of vehicle scheduling, and solves the problems of load imbalance and long path planning time in the prior art. Comprising the following steps: obtaining an initial population composed of N chromosomes; performing crossover, variation and error correction on each chromosome, and adding the chromosomes into the initial population to obtain a new population; calculating the fitness of each chromosome, selecting N chromosomes as an initial population, and carrying out iterative calculation again until a final population is obtained; obtaining a task allocation result of each platform from the final population; the path planning of one unmanned platform comprises the steps of obtaining obstacle information based on a running path of a planned vehicle, determining whether a path scheme for a node to arrive at a next adjacent node in the path planning has a conflict or not based on the obstacle information, and determining a next arriving node from adjacent nodes without the conflict. And rapid planning of load balancing is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicle scheduling, in particular to a multi-task scheduling method for multiple unmanned platforms. BACKGROUND

[0002] With the development of science and technology, multi-unmanned platform scheduling plays a vital role in many fields such as modern transportation, logistics, warehousing and automation. The unmanned vehicle platform scheduling system under multi-task can intelligently allocate resources according to task demand, vehicle state and traffic conditions, etc., to ensure efficient use of resources, reduce operating costs and improve economic efficiency.

[0003] Traditional scheduling methods cannot guarantee the balance of task allocation, and there is a situation of insufficient utilization of unmanned platform resources. In addition, with the increase in the number of unmanned vehicles and the amount of tasks, the complexity and optimization difficulty of the scheduling algorithm also increase, and the traditional scheduling algorithm has the problems of long planning time and repeated path planning when dealing with various complex traffic scenarios and task demands. SUMMARY

[0004] In view of the above analysis, the embodiments of the present application aim to provide a multi-task scheduling method for multiple unmanned platforms to solve the problems of uneven load and long path planning time in the prior art.

[0005] The embodiments of the present application provide a multi-task scheduling method for multiple unmanned platforms, comprising the following steps:

[0006] Based on the task information, a genetic algorithm is used to allocate tasks to each unmanned platform, including: based on the combination of task number and unmanned platform number, an initial population composed of N chromosomes is obtained; after crossing, mutating and error correcting each chromosome, the initial population is added to obtain a new population with a size of 2N; the fitness of each chromosome in the new population is calculated based on Manhattan distance, and N chromosomes are selected based on the fitness to be considered as an initial population for iterative calculation again until a final population is obtained; a chromosome is randomly selected from the final population to obtain the task allocation result of each platform;

[0007] Based on the task allocation result, the paths of each unmanned platform are planned; for path planning of one of the unmanned platforms, including: obtaining obstacle information based on the running path of the planned vehicle, determining whether there is a conflict in the path scheme from one node to the next adjacent node in the path planning based on the obstacle information, and determining the next arrival node from the adjacent nodes without conflict.

[0008] A further improvement to the above method involves obtaining a chromosome based on a combination of task number and unmanned platform number, including: randomly sorting all task numbers to obtain task codes; randomly selecting unmanned platform numbers equal to the number of tasks and sorting them to obtain unmanned platform codes; and concatenating the task codes and unmanned platform codes to obtain the chromosome.

[0009] Based on a further improvement of the above method, the step of selecting N chromosomes as the initial population based on fitness includes:

[0010] The steps for performing fast non-dominated sorting to obtain the first-level dominance layer include: placing the chromosome with the best fitness in the population into the dominance layer; removing the chromosome placed into the dominance layer from the population to obtain the population used to calculate the next level of dominance layer.

[0011] Constructing an initial population based on chromosomes in the dominance layer includes: sequentially adding chromosomes from the dominance layer to the initial population to be constructed in ascending order of dominance layer level, and calculating the number of chromosomes that can be accommodated in the initial population to be constructed; if the number of chromosomes that can be accommodated is less than the number of chromosomes in the dominance layer to be added, then randomly selecting chromosomes from that layer to add to the initial population to be constructed, so that the total number of chromosomes in the initial population is N.

[0012] Based on a further improvement of the above method, the steps for obtaining the chromosome with the best fitness in the population include: traversing and calculating the difference C1 between the first fitness C1 of one chromosome and another chromosome in the population. ′ The difference between the second fitness C2 and C2 ′ If C1 does not exist ′ and C2 ′ All are greater than 0, and there is no C1. ′ and C2 ′ If one is greater than 0, the other is equal to 0, and there exists at least one set of C1. ′ and C2 ′ All less than 0 or C1 ′ and C2 ′ If one chromosome is less than 0 and the other is equal to 0, then the chromosome with the best fitness is considered to be the chromosome with the best fitness.

[0013] Based on a further improvement of the above method, the first fitness C1 and the second fitness C2 of the chromosome are calculated using the following formula based on the Manhattan distance:

[0014]

[0015] Where n represents the number of unmanned platforms, m i Let x represent the number of tasks undertaken by the i-th unmanned platform, and j represent the j-th task of the i-th unmanned platform. ig,jy ig,j (x) represents the endpoint coordinates of the j-th task on the i-th unmanned platform. io,j y io,j (x) represents the starting coordinates of the i-th unmanned platform and the j-th task, (x) ci ,y c i) represents the coordinates of the node where the i-th unmanned platform is initially located; m average This indicates the average number of tasks undertaken by each unmanned platform.

[0016] Based on further improvements to the above method, the random selection of a chromosome from the final population to obtain the task allocation results for each platform includes: a one-to-one correspondence between the order of task numbers in the task code and the order of unmanned platform numbers in the unmanned platform code, indicating that the task is assigned to the corresponding unmanned platform; when an unmanned platform undertakes more than two tasks, the tasks are executed sequentially according to the order of arrangement.

[0017] Based on further improvements to the above method, the obstacle information includes the departure time and node of the unmanned platform, the arrival time and node of the destination, the running trajectory, the nodes along the route, and the time of arrival at each node along the route.

[0018] Based on a further improvement of the above method, the conflict includes point conflict and / or edge conflict; determining whether a point conflict exists in the path planning of a path from one node to the next adjacent node based on the obstacle information includes:

[0019] The distance between a node and its next adjacent node in the path scheme of the unmanned platform is obtained, along with the running speed, acceleration, and deceleration in the unmanned platform data matrix, to obtain the running time period; wherein, the distance is obtained from the map distance matrix;

[0020] By utilizing the departure time, arrival time, and arrival time of each unmanned platform in the obstacle information, the unmanned platforms with running trajectories in the obstacle information within the time period are obtained.

[0021] For each time point within the time period, calculate the distance between the center points of the unmanned platform and the unmanned platforms whose operating trajectories exist in the obstacle information; wherein, the distance is obtained using the nodes from which the unmanned platform departs, the nodes from which it reaches its destination, the operating trajectory, and the nodes along the route within the time period in the obstacle information;

[0022] Using the longest distance from the center point of the unmanned platform to the edge of the platform in the unmanned platform data matrix, calculate the sum of the longest distances between the two unmanned platforms;

[0023] A point conflict is considered to exist when the distance between the center points is less than or equal to the sum of the longest distance values.

[0024] Determining whether there are edge conflicts in the path planning from one node to the next adjacent node based on the obstacle information includes:

[0025] The unmanned platform obtains the road it travels from one node to the next according to the path scheme and the travel time of the travel on the road; wherein, the travel time is obtained by the distance between the starting and ending nodes and the operating speed, acceleration and deceleration of the unmanned platform;

[0026] By utilizing the departure time, arrival time, and arrival time of each unmanned platform in the obstacle information, the unmanned platforms with running trajectories in the obstacle information within the time period are obtained.

[0027] The path traveled by the unmanned platform is obtained from the nodes of departure, arrival, trajectory, and route of the unmanned platform within the time period specified in the obstacle information.

[0028] Determine whether the unmanned platform and the obstacle information within the time period enter the same road in opposite directions. If so, it is considered that there is an edge conflict.

[0029] Based on a further improvement to the above method, determining the next node from non-conflicting neighboring nodes includes:

[0030] Calculate the cost value of each adjacent node that does not conflict;

[0031] The point with the smallest cost value is selected as the next node; where the cost value represents the sum of the path length from the starting point to the node and the estimated path length from the node to the destination.

[0032] A further improvement to the above method is characterized by pre-modeling the map and unmanned platform that the task needs to run, thereby obtaining the distance matrix of the map and the data matrix of the unmanned platform; the distance matrix includes the distance between each node; the data matrix of the unmanned platform includes the number of the unmanned platform, its running speed, acceleration, deceleration, and the longest distance from the center point of the unmanned platform to the edge of the platform.

[0033] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0034] 1. This solution provides a multi-task scheduling method for multiple unmanned platforms. This method uses a genetic algorithm as its core. When faced with the situation of multiple tasks being issued, it aims to generate a task allocation scheme with the goal of unmanned platform load balancing and the shortest estimated total completion time.

[0035] 2. This scheme uses sequential planning as the main path planning principle to generate path planning schemes. When searching for paths, the time dimension is taken into account, avoiding collisions and conflicts in the scheduling process and realizing scheduling effectiveness. This method can realize collision-free path planning for multiple unmanned platforms under the condition of multiple task issuance.

[0036] 3. The path planning algorithm used in this solution reduces the complexity and time cost of the algorithm while ensuring the suboptimal nature of the solution, making it feasible to implement.

[0037] In this invention, the above-described technical solutions can be combined with each other to achieve more preferred combinations. Other features and advantages of this invention will be set forth in the following description, and some advantages may become apparent from the description or be learned by practicing the invention. The objects and other advantages of this invention can be realized and obtained from what is particularly pointed out in the description and drawings. Attached Figure Description

[0038] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts.

[0039] Figure 1 This is a flowchart illustrating a multi-task scheduling method for multiple unmanned platforms according to an embodiment of the present invention;

[0040] Figure 2 This is a schematic diagram of a chromosome obtained based on a combination of task number and unmanned platform number, according to an embodiment of the present invention.

[0041] Figure 3 This is a schematic diagram illustrating chromosome crossing according to an embodiment of the present invention;

[0042] Figure 4 This is a schematic diagram illustrating chromosome mutation according to an embodiment of the present invention;

[0043] Figure 5 This is a schematic diagram illustrating chromosome error correction according to an embodiment of the present invention;

[0044] Figure 6 This is a flowchart illustrating a path planning process for an unmanned platform based on task allocation results, as shown in one embodiment of the present invention. Detailed Implementation

[0045] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0046] A specific embodiment of the present invention discloses a multi-task scheduling method for multiple unmanned platforms, such as... Figure 1 As shown, it includes the following steps:

[0047] S11: Based on task information, a genetic algorithm is used to assign tasks to each unmanned platform, including: obtaining an initial population of N chromosomes based on the combination of task number and unmanned platform number; performing crossover, mutation, and error correction on each chromosome and adding it to the initial population to obtain a new population of size 2N; calculating the fitness of each chromosome in the new population based on Manhattan distance, selecting N chromosomes based on fitness as the initial population and iterating again until the final population is obtained; randomly selecting a chromosome from the final population to obtain the task allocation results for each platform.

[0048] S12: Plan the path of each unmanned platform based on the task allocation result; the path planning for one of the unmanned platforms includes: obtaining obstacle information based on the planned vehicle operation path, determining whether there is a conflict in the path plan of a node to the next adjacent node based on the obstacle information, and determining the next destination node from the adjacent nodes that do not have a conflict.

[0049] During implementation, the map and unmanned platform to be operated by the task are modeled in advance to obtain the distance matrix of the map and the data matrix of the unmanned platform. The distance matrix includes the distance between each node. The data matrix of the unmanned platform includes the number of the unmanned platform, its operating speed, acceleration, deceleration, and the longest distance from the center point of the unmanned platform to the edge of the platform.

[0050] Specifically, modeling the map required for the task involves: inputting the coordinates of each node in the task scenario map and the distances between each node, and outputting a distance matrix representing the distances between nodes. The data in the a-th row and b-th column of the distance matrix represents the distance between the a-th node and the b-th node. Here, a node refers to each location point in the task scenario involved in multiple tasks; for nodes that are not directly connected, the distance between nodes is set to infinity. For example, in a scenario where multiple unmanned platforms collaboratively load and unload goods in a warehouse, each loading and unloading point in the warehouse is set as a node, and by default, all paths are single-lane but allow two-way traffic.

[0051] Specifically, modeling the unmanned platforms involves taking as input the platform's ID, operating speed, acceleration, deceleration, and the longest distance from its center point to its edge. The output is a data matrix for each unmanned platform. In the data matrix, the i-th row represents the i-th unmanned platform, the first column is the platform's ID, the second column is its operating speed, the third column is its acceleration, the fourth column is its deceleration, and the fifth column is the longest distance from its center point to its edge.

[0052] like Figure 2 As shown, the steps for obtaining one chromosome based on the combination of task number and unmanned platform number include: randomly sorting all task numbers to obtain task codes; randomly selecting unmanned platform numbers equal to the number of tasks and sorting them to obtain unmanned platform codes; and concatenating the task codes and unmanned platform codes to obtain the chromosome. It should be noted that if the number of unmanned platforms is less than the number of tasks, the unmanned platform number can appear repeatedly in the unmanned platform code. It should also be noted that the initial number of chromosomes N in the population is an even number; for example, N is 100.

[0053] Furthermore, crossover, mutation, and error correction on each chromosome include: such as Figure 3 As shown, N chromosomes are grouped into pairs. The first S gene positions of the task code of the two chromosomes within a group are exchanged, and the first S gene positions of the unmanned platform code of the two chromosomes within a group are also exchanged, resulting in N exchanged chromosomes. For example, S is 3. Figure 4 As shown, the numbers of one gene locus in each chromosome obtained after the crossover mutation are randomly replaced to complete the mutation; the error correction operation refers to judging whether the chromosome after the crossover mutation has any unreasonable encoding, in combination with the actual application scenario. Figure 5 As shown, the specific operations include:

[0054] Iterate through the task codes to check for duplicate or missing task numbers;

[0055] If duplicate task numbers exist, delete the subsequent duplicate task numbers while retaining their gene loci; then, insert the missing task numbers into the empty gene loci in numerical order to obtain the corrected chromosome, and add it to the initial population to obtain a new population of size 2N. For example, if task numbers 7, 8, and 5 are duplicates, after deleting the subsequent duplicate task numbers, insert the missing task numbers 1, 3, and 6 in sequence to obtain the corrected task code.

[0056] Further, the fitness of each chromosome in the new population is calculated based on the Manhattan distance. Based on the fitness, N chromosomes are selected as the initial population and the calculation is repeated until the final population is obtained. The initial population is constructed based on the chromosomes in the dominance layer, including: adding the chromosomes in the dominance layer to the initial population to be constructed in ascending order of dominance layer level, and calculating the number of chromosomes that can be accommodated in the initial population to be constructed. If the number of chromosomes that can be accommodated is less than the number of chromosomes in the dominance layer to be added, then chromosomes are randomly selected from that layer and added to the initial population to be constructed, so that the total number of chromosomes in the initial population is N.

[0057] The step of selecting N chromosomes as the initial population based on fitness includes:

[0058] The steps for performing fast non-dominated sorting to obtain the first-level dominance layer include: placing the chromosome with the best fitness in the population into the dominance layer; removing the chromosome placed into the dominance layer from the population to obtain the population used to calculate the next level of dominance layer; wherein, the step of obtaining the chromosome with the best fitness in the population includes: traversing and calculating the difference C1 between the first fitness C1 of one chromosome and another chromosome in the population. ′ The difference between the second fitness C2 and C2 ′ If C1 does not exist ′ and C2 ′ All are greater than 0, and there is no C1. ′ and C2 ′ If one is greater than 0, the other is equal to 0, and there exists at least one set of C1. ′ and C2 ′ All less than 0 or C1 ′ and C2 ′ If one chromosome is less than 0 and the other is equal to 0, then the chromosome with the best fitness is considered to be the chromosome with the best fitness.

[0059] Specifically, using the distance matrix of the map and based on Manhattan distance, the first fitness C1 and the second fitness C2 of the chromosome are calculated using the following formula:

[0060]

[0061] Where n represents the number of unmanned platforms, m i Let x represent the number of tasks undertaken by the i-th unmanned platform, and j represent the j-th task of the i-th unmanned platform. ig,j y ig,j (x) represents the endpoint coordinates of the j-th task on the i-th unmanned platform. io,j y io,j (x) represents the starting coordinates of the i-th unmanned platform and the j-th task, (x) ci ,y c i) represents the coordinates of the node where the i-th unmanned platform is initially located; m average This indicates the average number of tasks undertaken by each unmanned platform.

[0062] Furthermore, a chromosome is randomly selected from the final population to obtain the task allocation results for each platform, including: a one-to-one correspondence between the order of task numbers in the task code and the order of unmanned platform numbers in the unmanned platform code, indicating that the task is assigned to the corresponding unmanned platform; when an unmanned platform undertakes more than two tasks, the tasks are executed sequentially according to the order of arrangement.

[0063] Next, based on the task allocation results, the paths for each unmanned platform will be planned, such as... Figure 6 As shown, the steps include:

[0064] Furthermore, obstacle information is obtained based on the planned path of the unmanned platform, including: the departure time and node of the unmanned platform, the arrival time and node of the destination, the running trajectory, the nodes along the route, and the time of arrival at each node along the route.

[0065] Path planning is performed on each unmanned platform in ascending order of their platform numbers. During path planning, the tasks assigned to each platform are executed sequentially according to their assigned order. Specifically, the path planning algorithm includes the following steps:

[0066] Step 1: Initialize the starting node as s and the target node as g, and put s into the set clo_set; set the variable current to the initial value of s, and the initial state of current as [Tc, localization], where Tc represents the time when the unmanned platform arrives at the current point, the initial value of Tc is the time when the platform is in its initial state, and localization represents the node currently set by current;

[0067] Step 2: Obtain the current neighboring nodes (including those waiting in place, i.e., the current point itself), and update the state of each neighboring point to [Tc+tn, localization], where tn is the estimated time for the unmanned platform to travel from the starting point to that point. If it is waiting in place, the estimated time is set to 1 second. The estimated time is calculated from the distance from the starting point to that point and the speed of the unmanned platform. The distance is obtained through the map matrix; the speed includes the acceleration, deceleration, and running speed recorded in the unmanned platform matrix.

[0068] Step 3: Iterate through the nodes adjacent to current in sequence and check whether there is a conflict between the path of the unmanned platform from current to the adjacent node and the path scheme in the obstacle; the conflict includes point conflict and edge conflict.

[0069] For example, determining whether there is a point conflict when the unmanned platform reaches the adjacent node from the current node includes:

[0070] The distance between a node and its next adjacent node in the path scheme of the unmanned platform is obtained, along with the running speed, acceleration, and deceleration in the unmanned platform data matrix, to obtain the running time period; wherein, the distance is obtained from the map distance matrix;

[0071] By using the departure time, arrival time, and arrival time of each unmanned platform in the obstacle to obtain the unmanned platforms with running trajectories in the obstacle during the time period;

[0072] For each time point within the time period, calculate the distance between the center points of the unmanned platform and the unmanned platforms with running trajectories in the obstacle; wherein, the distance is obtained using the nodes where the unmanned platform departs, the nodes where it reaches its destination, the running trajectory, and the nodes along the route within the time period in the obstacle;

[0073] Using the longest distance from the center point of the unmanned platform to the edge of the platform in the unmanned platform data matrix, calculate the sum of the longest distances between the two unmanned platforms;

[0074] A point conflict is considered to exist when the distance between the center points is less than or equal to the sum of the longest distance values; otherwise, a point conflict is considered not to exist.

[0075] For example, determining whether there is an edge conflict when the unmanned platform reaches the adjacent node from the current node includes:

[0076] The unmanned platform obtains the road it travels from one node to the next according to the path scheme and the travel time of the travel on the road; wherein, the travel time is obtained by the distance between the starting and ending nodes and the operating speed, acceleration and deceleration of the unmanned platform;

[0077] By using the departure time, arrival time, and arrival time of each unmanned platform in the obstacle to obtain the unmanned platforms with running trajectories in the obstacle during the time period;

[0078] The path traveled by the unmanned platform in the obstacle is obtained by using the nodes of departure, arrival, trajectory, and route of the unmanned platform within the time period specified in the obstacle.

[0079] Determine whether the unmanned platform and the unmanned platform in the obstacle enter the same road in opposite directions during the time period. If so, it is considered that there is an edge conflict; otherwise, it is considered that there is no edge conflict.

[0080] Step 4: Add neighboring nodes without any conflicts to the set neighbor_list, update the cost value of each node in neighbor_list, where the cost value represents the sum of the path length from the starting point to the point and the estimated path length from the point to the destination, and add the points in neighbor_list to ope_set.

[0081] Step 5: Determine if the end point has been found. If the end point has been found, end the current search; otherwise, select the point with the lowest cost in ope_set, add the point to clo_set and set current=point, and repeat steps 2 to 5.

[0082] Furthermore, the unmanned platform that has completed path planning is added to the close_set and removed from the open_set; since the planned path trajectory will be regarded as an obstacle by other vehicles, the calculated path of the unmanned platform is updated to the obstacle; the planned path is stored as a motion command.

[0083] If open-set is not empty, repeat steps 1 to 5 to plan the paths of the unmanned platforms in open_set; if open-set is empty, end the calculation, and the path planning for all unmanned platforms is now complete.

[0084] Next, the newly planned path will be sent to the corresponding unmanned platform in close_set through the scheduling system.

[0085] Compared with existing technologies, the solution provided in this embodiment offers a multi-task scheduling method for multiple unmanned platforms. This method uses a genetic algorithm as its core. When faced with multiple task assignments, it generates a task allocation scheme with the goals of unmanned platform load balancing and minimizing the estimated total completion time. After task allocation, it generates a path planning scheme using sequential planning as the main path planning principle. The time dimension is taken into account when searching for paths, avoiding collisions and conflicts during the scheduling process, thus achieving scheduling effectiveness and realizing collision-free path planning for multiple unmanned platforms under multiple task assignments. The path planning algorithm used reduces the complexity and time cost of the algorithm while ensuring the suboptimal nature of the solution, making it feasible for implementation.

[0086] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware, and the program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0087] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-task scheduling method for multiple unmanned platforms, characterized in that, The method comprises the following steps: The genetic algorithm is used to allocate tasks to each unmanned platform based on the task information, including: obtaining an initial population composed of N chromosomes based on the combination of task numbers and unmanned platform numbers; adding each chromosome to the initial population after performing cross, mutation and error correction to obtain a new population with a size of 2N; calculating the fitness of each chromosome in the new population based on the Manhattan distance, and selecting N chromosomes with the highest fitness as the initial population for iterative calculation until a final population is obtained; and selecting a chromosome from the final population at random to obtain the task allocation result of each platform. The paths of the unmanned platforms are planned based on the task allocation result, and the path planning of one of the unmanned platforms comprises: obtaining obstacle information based on the running path of the vehicle, determining whether there is a conflict in the path scheme of one node to the next adjacent node in the path planning based on the obstacle information, and determining the next arrival node from the adjacent nodes without conflict.

2. The multi-task scheduling method for multiple unmanned platforms according to claim 1, wherein, The step of obtaining one chromosome based on the combination of task numbers and unmanned platform numbers comprises: randomly sorting all task numbers to obtain a task code, randomly selecting unmanned platform numbers equal to the number of tasks and sorting them to obtain an unmanned platform code, and splicing the task code and the unmanned platform code to obtain the chromosome.

3. The method of claim 2, wherein, The step of selecting N chromosomes with the highest fitness as the initial population based on the fitness comprises: The step of performing fast non-dominated sorting to obtain a first-level dominated layer comprises: placing the chromosome with the highest fitness in the population into the dominated layer; and deleting the chromosome placed in the dominated layer from the population to obtain a population for calculating the next-level dominated layer; The step of constructing an initial population based on the chromosomes in the dominated layer comprises: adding the chromosomes in the dominated layer to the initial population to be constructed in order from small to large according to the level of the dominated layer, and calculating the number of remaining accommodated chromosomes in the initial population to be constructed; if the number of remaining accommodated chromosomes is less than the number of chromosomes in the dominated layer to be added, randomly selecting chromosomes from the layer to add to the initial population to be constructed, so that the total number of chromosomes in the initial population is N.

4. The method of claim 3, wherein, The step of obtaining the chromosome with the best fitness in the population comprises: traversing the difference C1 between the first fitness C1 of a chromosome in the population and the first fitness C1 of another chromosome in the population ′ and the difference C2 between the second fitness C2 of the chromosome and the second fitness C2 of the other chromosome ′ If C1 ′ and C2 ′ are both greater than 0, and C1 ′ and C2 ′ are not both greater than 0, and at least one of C1 ′ and C2 ′ is less than 0, or one of C1 ′ and C2 ′ is less than 0 and the other is equal to 0, then the chromosome is considered to be the chromosome with the best fitness.

5. The multi-task scheduling method for multiple unmanned platforms according to claim 4, wherein, The first fitness C1 and the second fitness C2 of the chromosome are calculated based on the Manhattan distance by using the following formula: wherein n represents the number of unmanned platforms, m i represents the number of tasks undertaken by the i-th unmanned platform, j represents the j-th task of the i-th unmanned platform, (x ig,j , y ig,j ) represents the end point coordinates of the j-th task of the i-th unmanned platform, (x io,j , y io,j ) represents the start point coordinates of the j-th task of the i-th unmanned platform, (x ci , y c i) represents the coordinates of the node where the initial state of the i-th unmanned platform is located; m average represents the average number of tasks undertaken by each unmanned platform.

6. The multi-task scheduling method for multiple unmanned platforms according to claim 5, wherein, The step of selecting a chromosome from the final population at random to obtain the task allocation result of each platform comprises: the arrangement order of the task numbers in the task code and the arrangement order of the unmanned platform numbers in the unmanned platform code correspond one-to-one, indicating that the task is allocated to the corresponding unmanned platform; when one unmanned platform undertakes more than two tasks, the tasks are executed in turn according to the arrangement order.

7. The multi-task scheduling method for multiple unmanned platforms according to claim 1, wherein, The obstacle information comprises the time point and node of the departure of the unmanned platform, the time point and node of the arrival at the terminal, the running trajectory, the nodes of the route and the arrival time at each node of the route.

8. The multi-task scheduling method for multiple unmanned platforms according to claim 7, wherein, The conflict comprises point conflict and / or edge conflict. The step of determining whether there is a point conflict in the path scheme of one node to the next adjacent node in the path planning based on the obstacle information comprises: obtain a running time period by acquiring a distance between a node and a next adjacent node in the path scheme of the unmanned platform and a running speed, an acceleration and a deceleration in the unmanned platform data matrix; wherein the distance is obtained from a map distance matrix; obtain the unmanned platforms with running trajectories in the obstacle information in the time period by using the time points at which the unmanned platforms in the obstacle information start, the time points at which the unmanned platforms reach the end point and the time points at which the unmanned platforms reach the nodes on the way; calculate the center point distances between the unmanned platform and the unmanned platforms with running trajectories in the obstacle information at each time point in the time period; wherein the distances are obtained by using the nodes at which the unmanned platforms in the obstacle information start, the nodes at which the unmanned platforms reach the end point, the running trajectories and the nodes on the way in the obstacle information in the time period; calculate the sum of the longest distance values of the two unmanned platforms by using the longest distance value from the center point of the unmanned platform to the edge of the platform in the unmanned platform data matrix; consider that there is a point conflict when the center point distance is less than or equal to the sum of the longest distance values; determine whether there is an edge conflict in the path scheme from a node to a next adjacent node in the path planning based on the obstacle information, comprising: obtain a road on which the unmanned platform travels from a node to a next node according to the path scheme and a running time period of the road; wherein the running time period is obtained by a distance between the start node and the end node and a running speed, an acceleration and a deceleration of the unmanned platform; obtain the unmanned platforms with running trajectories in the obstacle information in the time period by using the time points at which the unmanned platforms in the obstacle information start, the time points at which the unmanned platforms reach the end point and the time points at which the unmanned platforms reach the nodes on the way; obtain the road on which the unmanned platforms in the obstacle information travel by using the nodes at which the unmanned platforms in the obstacle information start, the nodes at which the unmanned platforms reach the end point, the running trajectories and the nodes on the way in the obstacle information in the time period; determine whether the unmanned platform and the unmanned platforms in the obstacle information enter the same road in opposite directions in the time period, and if so, consider that there is an edge conflict.

9. The method of claim 1, wherein, the next node is determined from the adjacent nodes without conflict, comprising: calculate the cost values of the adjacent nodes without conflict; select the node with the minimum cost value as the next node; wherein the cost value represents the sum of the length of the path from the start node to the node and the length of the estimated path from the node to the end point.

10. The multi-task scheduling method for multiple unmanned platforms according to any one of claims 1, 7, 8 or 9, wherein, model the map on which the task needs to run and the unmanned platform in advance to obtain a distance matrix of the map and a data matrix of the unmanned platform; the distance matrix comprises distances between nodes; and the data matrix of the unmanned platform comprises a number of the unmanned platform, a running speed, an acceleration, a deceleration and a longest distance value from the center point of the unmanned platform to the edge of the platform.

Images