A spatiotemporal collaborative task allocation method for heterogeneous multi-unmanned platforms

Abstract

The present invention discloses a method for allocating heterogeneous multi-unmanned platform spatiotemporal collaborative tasks, which includes: step S1: reading and storing task data to be executed; step S2: dividing the loads required for the task into different categories; step S3: inputting various attribute data of ground unmanned platforms and aerial unmanned platforms that can execute tasks; step S4: making improvements to the ant colony algorithm by introducing time penalties and / or ranking incentives; step S5: matching demand information, and using the improved ant colony algorithm to maximize the efficiency of unmanned platforms and tasks to obtain the optimal task allocation plan; step S6: according to the changes in tasks and the dynamic update of the status of unmanned platforms, real-time calculation and adjustment of the execution plan until all tasks to be executed are completed. The present invention has the advantages of better flexibility, higher intelligence, faster calculation efficiency, etc.

Description

Technical Field

[0001] The present invention mainly relates to the technical field of intelligent unmanned equipment control, and in particular to a method for spatiotemporal collaborative task allocation of heterogeneous multi-unmanned platforms based on an improved ant colony algorithm. Background Art

[0002] With the development of science and technology, more and more intelligent unmanned equipment are being used in all walks of life. For example, there are many intelligent unmanned equipment (unmanned systems) such as drones and unmanned vehicles. Based on the known task sequence, who will perform each action, when to perform it, how to meet the task requirements as much as possible, how to improve the efficiency of the action, etc., these are all urgent problems to be solved in task allocation. Due to the heterogeneity of unmanned platforms, their load capacity, moving speed, moving routes, etc. are all different. In addition, there are logical constraints and time window requirements for tasks. Traditional task allocation methods are difficult to meet actual needs.

[0003] In the context of complex tasks, for a given number of unmanned platforms that can carry different payloads and have different capabilities, how to achieve the optimization goal of meeting the time window requirements and minimizing the total action time while satisfying the capability requirements of each task point and the constraints of orderly actions for the task sequence to be executed is an urgent problem to be solved in the spatiotemporal collaborative task allocation of heterogeneous multi-unmanned platforms. Summary of the invention

[0004] The technical problem to be solved by the present invention is: in response to the technical problems existing in the prior art, the present invention provides a method for allocating spatiotemporal collaborative tasks of heterogeneous multi-unmanned platforms with better flexibility and higher intelligence.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0006] A method for allocating spatiotemporal collaborative tasks among heterogeneous multi-unmanned platforms, comprising the following steps:

[0007] Step S1: Read and store the task data to be executed;

[0008] Step S2: Divide the loads required for the task into different categories;

[0009] Step S3: inputting various attribute data of the ground unmanned platform and the aerial unmanned platform that can perform the task;

[0010] Step S4: Improve the ant colony algorithm by introducing time penalty and / or ranking incentive;

[0011] Step S5: Match the demand information and use the improved ant colony algorithm to maximize the efficiency of pairing the unmanned platform and the task to obtain the optimal task allocation plan;

[0012] Step S6: Calculate and adjust the execution plan in real time according to the changes in tasks and the dynamic update of the status of the unmanned platform until all tasks to be executed are completed.

[0013] As a further improvement of the method of the present invention: the task data includes one or more of a task sequence, a task time window, a load requirement, a water bomb quantity requirement, current map information, and road passability.

[0014] As a further improvement of the method of the present invention: the load corresponds to one or more tasks.

[0015] As a further improvement of the method of the present invention: the time windows of different tasks in step S1 include hard time windows that must be met and soft time windows that are met as much as possible.

[0016] As a further improvement of the method of the present invention: in the step S2, the categories of the payload classification include one or more of reconnaissance payload, laser payload, and communication payload.

[0017] As a further improvement of the method of the present invention: in step S3, the attribute data of the ground unmanned platform and the aerial unmanned platform include one or more of power, moving speed, longitude and latitude position, carried load, and carried water bomb amount base.

[0018] As a further improvement of the method of the present invention: in the step S4, the ant colony algorithm is improved by introducing a time penalty, specifically, the heuristic item information of the time penalty is optimized; when searching for executable tasks, each unmanned platform, in addition to considering the load demand matching and the moving distance, must also combine the heterogeneous characteristics, calculate its arrival time according to the moving speed and road length of the unmanned platform, and calculate the difference between the arrival time and the estimated start time of the task, that is, the timeout time; if the unmanned platform arrives before the estimated start time of the task, no penalty is imposed; if the arrival time exceeds the timeout, the percentage of the difference between the arrival time and the shortest timeout time as the maximum difference is calculated, and a coefficient factor is attached to the heuristic item information.

[0019] As a further improvement of the method of the present invention: in the spatiotemporal collaborative task allocation model of heterogeneous multi-unmanned platforms, when the unmanned platform is looking for the next path, the heuristic term in the probability of the ant selecting the path from i to j is as shown in the following formula:

[0020]

[0021]

[0022] The heuristic term considers the moving distance and waiting time. In the formula, ω1 and ω2 represent the weight factors of the moving distance and waiting time respectively, d(i,j) represents the length of the path from i to j, and d min represents the shortest path among the optional paths, dmax represents the longest path among the optional paths, so C dis (i, j) represents the percentage of the difference between the path from i to j and the shortest path in the maximum difference; represents the start time of the action window of task point j, t(i,j) represents the arrival time after selecting the path from i to j, Represents the waiting time for selecting this path to execute the task, C win (i, j) represents the percentage of the waiting time of selecting path i to j to execute the task to the maximum waiting time.

[0023] As a further improvement of the method of the present invention: in the step S4, the ant colony algorithm is improved by introducing ranking incentives, specifically, the pheromone update optimization of ranking incentives; the objective function values ​​found by each ant are sorted from small to large, and only the top 20% of the ants are allowed to leave pheromones, and the weight is inversely proportional to the path length and multiplied by a ranking factor σ that decreases linearly with the ranking, that is, the ranking factor of the first place is 1, and the ranking factor at 20% is reduced to 0; the pheromone update optimization is shown in the following formula:

[0024]

[0025]

[0026] Among them, τ ij (t+1) represents the concentration of pheromone at time t+1, and the volatility factor ρ ranges from 0 to 1. represents the sum of pheromones updated by all ants from i to j, d ij represents the path length from i to j, σ k represents the ranking found by ant k based on the objective function, σ k The value is set to 1 for the first value and decreases to 0 at 20%.

[0027] Compared with the prior art, the advantages of the present invention are:

[0028] 1. The spatiotemporal collaborative task allocation method for heterogeneous multi-unmanned platforms of the present invention introduces time penalties and ranking incentives to improve the ant colony algorithm in view of the particularity and multiple constraints of the task allocation for heterogeneous multi-unmanned platforms. The present invention can reduce the waiting time of the unmanned platform and improve the search efficiency by introducing time penalties, which is conducive to finding the optimal path, reducing the total time of action, and completing the task more efficiently and quickly. The present invention increases the voice of better path finders by introducing ranking incentives, so that advanced experience can be promoted more smoothly, and backward experience is less likely to form self-locking and suppress the generation of advanced experience. It can be seen that the improvement of the algorithm by the present invention makes the overall task allocation more efficient and intelligent.

[0029] 2. The spatiotemporal collaborative task allocation method for heterogeneous multi-unmanned platforms of the present invention takes into account the heterogeneity, moving speed, carrying load, water bomb base, time window and other constraints of the unmanned platforms, and is closer to the actual situation; the ground unmanned platform can only move along the known roads, and the aerial unmanned platform can pass freely without constraints, which is basically consistent with the actual situation. It can be seen that the present invention can realize the spatiotemporal collaborative task allocation of heterogeneous multi-unmanned platforms in dynamic and complex scenes, and has good scientific research and engineering value. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a schematic diagram of the process of the present invention.

[0031] Figure 2 It is an improved sub-flow chart of introducing time penalty in the method of the present invention.

[0032] Figure 3 It is an improved sub-flow chart of introducing ranking incentive in the method of the present invention.

[0033] Figure 4 It is a schematic diagram comparing the effects of the improved ant colony algorithm and the traditional ant colony algorithm in the method of the present invention.

[0034] Figure 5 It is a schematic diagram of a disaster area situation map in a specific application example of the present invention.

[0035] Figure 6 It is a schematic diagram of a task sequence to be executed in a specific application example of the present invention.

[0036] Figure 7 It is a schematic diagram of the load carrying situation and regional coordinate situation of the unmanned platform in a specific application example of the present invention.

[0037] Figure 8 It is a schematic diagram of the test results of the present invention in a specific application example. Specific embodiments

[0038] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0039] like Figure 1-Figure 4 As shown, a method for allocating spatiotemporal collaborative tasks of heterogeneous multi-unmanned platforms of the present invention is actually a method for allocating spatiotemporal collaborative tasks of heterogeneous multi-unmanned platforms based on an improved ant colony algorithm; the steps of the present invention include:

[0040] Step S1: Read and store the task data to be executed;

[0041] In a specific application example, the mission data includes but is not limited to mission sequence, mission time window, load requirement, water bomb quantity requirement, current map information, road accessibility and other data.

[0042] Step S2: Divide the loads required for the task into different categories;

[0043] Specifically, when applied, one load can correspond to multiple tasks.

[0044] Step S3: inputting various attribute data of the ground unmanned platform and the aerial unmanned platform that can perform the task;

[0045] Step S4: Improve the ant colony algorithm by introducing time penalty and / or ranking incentive;

[0046] Step S5: matching demand information (such as load demand and water bomb quantity demand, etc.), and using the improved ant colony algorithm to maximize the efficiency of pairing the unmanned platform and the task to obtain the optimal task allocation plan;

[0047] Step S6: Calculate and adjust the execution plan in real time according to the changes in tasks and the dynamic update of the status of the unmanned platform until all tasks to be executed are completed.

[0048] In a specific application example, the time windows of different tasks in step S1 include a hard time window that must be met and a soft time window that is met as much as possible.

[0049] In a specific application example, in step S2, the categories of the payload classification can be adjusted according to actual conditions, such as reconnaissance payload, laser payload, communication payload, etc.

[0050] In a specific application example, in step S3, the attribute data of the ground unmanned platform and the aerial unmanned platform include but are not limited to some or all of the data such as power level, moving speed, longitude and latitude position, carried load, and water bomb amount carried.

[0051] See also Figure 2 In a specific application example, in step S4, the ant colony algorithm is improved by introducing a time penalty, specifically, the heuristic item information of the time penalty is optimized; when searching for executable tasks, each unmanned platform, in addition to considering the load demand matching and moving distance, must also combine heterogeneous characteristics, calculate its arrival time according to the moving speed of the unmanned platform, the length of the road, etc., and calculate the difference between the arrival time and the expected start time of the task, which is the timeout time. If the unmanned platform arrives before the expected start time of the task, there is no penalty; if the arrival time is overdue, the percentage of the difference between the shortest timeout time and the maximum difference is calculated, and a coefficient factor is added to the heuristic item information. The purpose of this is to hope that when the unmanned platform searches for the task path, it can not only consider the distance factor, but also meet the task time window requirements as much as possible, so as to maximize the execution efficiency of the collaborative task. By adjusting the coefficient factors of the two, the allocation needs of different purposes can be met.

[0052] The traditional ant colony algorithm uses η ij This value is used as the heuristic item for path i to j. The value of the heuristic item is inversely proportional to the distance from i to j. When the ant is looking for the next path, it will only decide based on the distance, which is easy to fall into the local optimum. In the spatiotemporal collaborative task allocation model of heterogeneous multi-unmanned platforms, when the unmanned platform is looking for the next path, it must consider not only the distance factor but also the time factor. Therefore, the heuristic item in the probability of the ant choosing path i to j is modified as shown in formula (1):

[0053]

[0054]

[0055] The heuristic term is changed from only considering the distance to considering the organic combination of moving distance and waiting time. In the formula, ω1 and ω2 represent the weight factors of moving distance and waiting time respectively, d(i,j) represents the length of path i to j, and d min represents the shortest path among the optional paths, d max represents the longest path among the optional paths, so C dis (i, j) represents the percentage of the difference between the path from i to j and the shortest path in the maximum difference.

[0056] Similarly, represents the start time of the action window of task point j, t(i,j) represents the arrival time after selecting the path from i to j, represents the waiting time for selecting this path to execute the task, so C win (i, j) represents the percentage of the waiting time of selecting path i to j to execute the task to the maximum waiting time.

[0057] The present invention adds the percentage of the difference between the selected timeout time and the shortest timeout time in the heuristic item information, and optimizes the distance factor to the percentage of the difference between the selected i to j path and the shortest path in the maximum difference. This can reduce the waiting time of the unmanned platform, improve the search efficiency, facilitate finding the optimal path, reduce the total time of the action, and complete the task more efficiently and quickly.

[0058] See also Figure 3 In a specific application example, in step S4, the ant colony algorithm is improved by introducing ranking incentives, specifically, pheromone update optimization of ranking incentives.

[0059] In the traditional ant colony algorithm, all ants have the same α and β parameters, which makes the algorithm's execution effect sensitively dependent on the settings of the parameters α and β. In addition, all ants leave pheromones, but the amount of pheromones left is inversely proportional to the length of their respective TSP paths, which makes the map full of ant "footprints" and makes the ants lose their sense of direction. It is very likely that an ant happens to walk out of the shortest TSP path, but is drowned in the chaotic pheromone signals and is finally forgotten by the entire ant colony. To avoid this situation, the traditional ant colony algorithm can only increase α, which is undoubtedly more likely to fall into a local minimum, reducing the possibility of another shorter TSP path being discovered later.

[0060] For example, in a collaborative search and rescue mission, the unmanned platform will search for action routes based on the decomposed task sequence and various constraints. In this process, if the result of a certain generation happens to have the best objective function value, it is necessary to give a large weight to the best result in time to enhance the voice of excellent results, and then continue to promote the advanced on the basis of the advanced, so that the excellent individuals can be even better, which greatly reduces the number of iterations for solving the entire problem and can converge to the global optimal solution faster.

[0061] Therefore, the traditional pheromone recording rules are optimized, and the objective function values ​​found by each ant are sorted from small to large. Only the top 20% of ants are allowed to leave pheromones. The weight is inversely proportional to the path length and multiplied by a factor σ (i.e., ranking factor) that decreases linearly with the ranking. That is, the ranking factor of the first ant is 1, and the ranking factor of the 20% ant is reduced to 0. The pheromone update optimization is shown in formula (2):

[0062]

[0063]

[0064] Among them, τ ij (t+1) represents the concentration of pheromone at time t+1, and the volatility factor ρ ranges from 0 to 1. represents the sum of pheromones updated by all ants from i to j, d ij represents the path length from i to j, σ k represents the ranking found by ant k based on the objective function, σ k The value is set to 1 for the first value and decreases to 0 at 20%.

[0065] In addition to being able to selectively promote successful experiences, the introduction of ranking factors in the present invention also increases the voice of those who discover better paths. Take 100 ants as an example. If more than half of the ants in the entire population are stuck on a suboptimal path, and one ant happens to find a shorter path, according to traditional rules, it only has 1 / 100 of the voice. However, if the ranking factor rules are followed, the weights of the top 20 ants decrease linearly, and the first ant has about 1 / 10 of the voice. This allows advanced experience to be promoted more smoothly, and backward experience is less likely to form self-locking and suppress the generation of advanced experience.

[0066] Taking a specific application as an example, the search area is set to a range of 2000m*2000m. The disaster area situation map is as follows: Figure 5 As shown, the task sequence to be executed is as follows Figure 6 As shown. During the search and rescue operation, each team carries 2 drones and 1 unmanned vehicle, and the initial positions of the unmanned platforms are consistent with the initial positions of their respective teams. The drone has a speed of 900m / min and carries 1 base of water bombs; the unmanned vehicle has a speed of 300m / min and carries 10 bases of water bombs. The task of extinguishing the fire point requires 1 base of ammunition, and the water bomb support task requires 3 bases of ammunition. Taking the current search and rescue start time as the reference zero point, each task is scheduled to start at the same time after 5 minutes. The load carrying situation and regional coordinates of each unmanned platform are shown as follows. Figure 7 shown.

[0067] The improved ant colony algorithm is used for calculation. The number of ants is set to 60, the number of iterations is 500, the values ​​of ω1 and ω2 in the heuristic information are both set to 0.5, the weighted value of pheromone α=1, the weighted value of visibility β=3, the volatility factor ρ=0.5, and the penalty factor γ in the objective function is set to 1. The test results are as follows: Figure 8 shown.

[0068] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A method for allocating tasks in time and space for heterogeneous multi-unmanned platforms, characterized in that the steps include: Step S1: Read and store the task data to be executed; Step S2: Divide the loads required for the task into different categories; Step S3: inputting various attribute data of the ground unmanned platform and the aerial unmanned platform that can perform the task; Step S4: Improve the ant colony algorithm by introducing time penalty and / or ranking incentive; Step S5: Match the demand information and use the improved ant colony algorithm to maximize the efficiency of pairing the unmanned platform and the task to obtain the optimal task allocation plan; Step S6: Calculate and adjust the execution plan in real time according to the changes in the tasks and the dynamic update of the status of the unmanned platform until all the tasks to be executed are completed; In step S4, time penalty improvement is performed in the ant colony algorithm, that is, the heuristic item information of time penalty is optimized; when searching for executable tasks, each unmanned platform considers load demand matching and moving distance, combines heterogeneous characteristics, calculates arrival time according to the moving speed and road length of the unmanned platform, and calculates the difference between the arrival time and the estimated start time of the task, that is, the timeout time; if the unmanned platform arrives before the estimated start time of the task, no penalty is imposed; if the arrival time is overdue, the percentage of the difference between the arrival time and the shortest timeout time to the maximum difference is calculated, and a coefficient factor is added to the heuristic item information; Improve the ranking incentive in the ant colony algorithm, that is, optimize the pheromone update for ranking incentive; sort the objective function values ​​found by each ant from small to large, and only allow the top 20% of ants to leave pheromones. The weight is inversely proportional to the path length and multiplied by a ranking factor that decreases linearly with the ranking. , that is, the ranking factor of the first place is 1, and the ranking factor is reduced to 0 at 20%; the pheromone update optimization is shown as follows: in, Represents the concentration of pheromone at time t+1, volatility factor The value range is from 0 to 1. Indicates passing arrive The sum of pheromones updated by all ants, represent arrive Path length, represents the ranking found by ant k based on the objective function, The value is set to 1 for the first value and decreases to 0 at 20%.

2. The method for spatiotemporal collaborative task allocation of heterogeneous multi-unmanned platforms according to claim 1 is characterized in that: The mission data includes one or more of a mission sequence, a mission time window, a load requirement, a water bomb requirement, current map information, and road passability.

3. The method for spatiotemporal collaborative task allocation of heterogeneous multi-unmanned platforms according to claim 2 is characterized in that: The payload corresponds to one or more tasks.

4. The method for spatiotemporal collaborative task allocation of heterogeneous multi-unmanned platforms according to claim 1 is characterized in that: The time windows of different tasks in step S1 include hard time windows that must be met and soft time windows that must be met.

5. The method for spatiotemporal collaborative task allocation of heterogeneous multi-unmanned platforms according to claim 1 is characterized in that: In step S2, the categories of the payload classification include one or more of reconnaissance payload, laser payload, and communication payload.

6. The method for allocating tasks of heterogeneous multi-unmanned platforms in time and space according to claim 1 is characterized in that: In step S3, the attribute data of the ground unmanned platform and the aerial unmanned platform include one or more of power, moving speed, longitude and latitude position, carried load, and carried water bomb amount base.

7. The method for spatiotemporal collaborative task allocation of heterogeneous multi-unmanned platforms according to claim 1 is characterized in that: In the spatiotemporal collaborative task allocation model of heterogeneous multi-unmanned platforms, when the unmanned platform is looking for the next path, the ants will select the path. arrive The heuristic term in the probability of is shown as follows: The heuristic term considers the moving distance and waiting time, where and Represent the weight factors of moving distance and waiting time respectively, Representative path arrive Length, Represents the shortest path among the optional paths, represents the longest path among the optional paths, so, Representative Selection arrive The difference between the path and the shortest path is the percentage of the maximum difference; represent The start time of the mission point action window, Representative Selection arrive The arrival time after this path, Represents the waiting time for selecting this route to execute the task. Representative Selection arrive The waiting time of the path execution task is the percentage of the maximum waiting time.

Images