Multi-angle cooperative attack task heterogeneous unmanned aerial vehicle cluster path planning method

Abstract

The application discloses a heterogeneous unmanned aerial vehicle cluster path planning method for multi-angle cooperative attack tasks, in which map initialization and unmanned aerial vehicle task allocation are firstly performed, then the path of each unmanned aerial vehicle is planned according to the kinematic characteristics and dynamic obstacles of the unmanned aerial vehicle and Dubins path constraints are considered, and finally multi-angle cooperative attack of the unmanned aerial vehicle is realized. The application also relates to three improvements of the A* algorithm, including automatic adjustment of the grid size, path smoothing processing and modification of the heuristic function. The application realizes heterogeneous unmanned aerial vehicle cluster path planning and task cooperation for multi-angle cooperative attack tasks, improves task efficiency and safety, and can adapt to task requirements in complex environments. The application field of the technology relates to military, security and commercial fields, and can improve target attack efficiency and reduce losses.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of aircraft path planning, in particular to a heterogeneous unmanned aerial vehicle cluster path planning method for multi-angle cooperative attack tasks. BACKGROUND

[0002] With the development of unmanned aerial vehicle technology, unmanned aerial vehicles are widely used in various fields, including military, civilian and commercial aspects. In cooperative attack tasks, unmanned aerial vehicles need to consider the defense situation and interaction of each target according to the requirements of cooperative attack, and perform path planning and task allocation according to task requirements, so as to realize the cooperation of multiple unmanned aerial vehicles and complete the attack of unified targets.

[0003] However, for multi-target cooperative attack tasks, there are many challenges in unmanned aerial vehicle path planning. Each unmanned aerial vehicle works on different targets and needs to dynamically allocate tasks according to task requirements; complex terrain, changing no-fly zones and weather and other external environmental factors in path planning will affect the flight trajectory of unmanned aerial vehicles, making it difficult for path planning algorithms to meet task requirements; in multi-target cooperative attack tasks, each unmanned aerial vehicle needs to adjust the path planning in interaction and cooperation to prevent collisions.

[0004] For existing multi-angle cooperative attack tasks of unmanned aerial vehicles, the kinematic characteristics of unmanned aerial vehicles and no-fly zones and other factors have an impact on path planning, making it difficult for path planning algorithms to simultaneously meet flight safety, efficiency and task requirements. Therefore, the present application aims to provide a heterogeneous unmanned aerial vehicle cluster path planning method for multi-angle cooperative attack tasks, which comprehensively considers the kinematic characteristics and no-fly zones and other factors in path planning to achieve multi-angle cooperative attack on targets. SUMMARY

[0005] The present application provides a heterogeneous unmanned aerial vehicle cluster path planning method for multi-angle cooperative attack tasks, which considers dynamic detection of no-fly zones, kinematic characteristics of unmanned aerial vehicles and other factors through map initialization, task allocation, path planning and local path planning, and achieves multi-angle cooperative attack on multiple attack targets.

[0006] A heterogeneous unmanned aerial vehicle cluster path planning method for multi-angle cooperative attack tasks, comprising the following steps:

[0007] Step 1: information initialization, mainly including: unmanned aerial vehicle and payload information initialization, terrain initialization, target initialization and enemy radar initialization;

[0008] Step 2: task allocation for multi-angle cooperative attack on multiple attack targets according to the initialized information;

[0009] Step 3: Path planning using improved A* algorithm, taking into account the kinematics of the UAV, and finally converting the path to Dubins path;

[0010] First, set the position of the UAV in the task chain as the starting point, set the strike point as the end point, and perform grid operation on the map, taking the starting point and end point as the diagonal endpoints of the grid map; Then use A* algorithm to find the optimal path in the grid map;

[0011] The heuristic function h(n) used in the improved A* algorithm is designed as follows:

[0012] f(n)=g(n)+h(n)

[0013]

[0014] Where f(n) is the evaluation value of the priority of each grid point n, g(n) is the cost of grid point n from the starting point, dis represents the Manhattan distance of grid point n from the end point, and the penalty term for obstacles is:

[0015]

[0016]

[0017] Where, represents the Euclidean distance between grid point n and the i-th obstacle vertex, R min represents the minimum hovering radius of the current UAV;

[0018] The penalty term for enemy radar is:

[0019]

[0020]

[0021] Where, represents the Euclidean distance between grid point n and the j-th enemy radar, represents the detection radius of the j-th enemy radar;

[0022] The penalty term for other paths is:

[0023]

[0024]

[0025] Where, represents the Euclidean distance between grid point n and the k-th other path point, dis' represents the Euclidean distance of grid point n from the end point;

[0026] Step 4: The UAVs start to perform the task according to the respective Dubins path;

[0027] Step 5: When the no-fly zone changes, check whether the path is affected, if affected, then perform local path planning to adjust the path;

[0028] Step 6: All UAVs complete the strike task, and the task ends.

[0029] Preferably, in the improved A* algorithm, the grid size is adjusted as needed to ensure that the starting point and the target point of the path planning are at the endpoints of the grid.

[0030] Preferably, the grid map is set as a 40*40 rectangular grid, and then 20 grids are expanded outward on this basis, finally obtaining an 80*80 rectangular grid map, and the starting point and the end point are just on the grid points.

[0031] Preferably, in the improved A* algorithm, when the Dubins path is calculated, the generated path is smoothed and the redundant inflection points are deleted.

[0032] Preferably, the information includes UAV and load information, terrain information, target information, and enemy radar information;

[0033] The UAV information includes UAV number, type, latitude and longitude, heading angle, minimum speed, and maximum speed;

[0034] The UAV type includes fixed-wing UAV and rotary-wing UAV;

[0035] The load information includes attack load number, strike mode, default launch distance, and number of ammunition;

[0036] The load strike mode includes guided and unguided;

[0037] The terrain information includes latitude and longitude information of each vertex of the fixed obstacle, time windowed no-fly zone no-fly time range, and latitude and longitude information of each vertex of the no-fly zone;

[0038] The target information includes target point latitude and longitude information, number of aircrafts that need to be cooperatively attacked, and strike direction angle;

[0039] The enemy radar information includes latitude and longitude information of the enemy radar and detection radius of the radar.

[0040] Preferably, firstly, the number of unmanned planes required to cooperatively attack each target is determined, and the unmanned plane that best matches the current task is calculated according to an evaluation function; then, the attack point position and attack direction of each unmanned plane are calculated according to the default attack distance of the target point information and the load information; the process is repeated until the task allocation for all target points is completed.

[0041] Preferably, the evaluation function used is as follows:

[0042] Value = a · (dis - dis atack ) + b · (|heading uas - heading attack |)

[0043] Wherein, a and b are evaluation coefficients of distance and angle respectively, dis is the Euclidean distance of the unmanned plane from the target point, dis atack is the default launch distance of the attack load, heading uas is the initial direction angle of the unmanned plane, heading attack is the attack direction angle of the target, and heading is the direction angle of the unmanned plane pointing to the target point, and a and b are set coefficients.

[0044] Preferably, a is 0.001 and b is 57.296.

[0045] Preferably, the local path planning process comprises: firstly, an initial path is calculated using an improved A* algorithm according to the current position and the target position; then, in the process of local path planning, it is constantly checked whether the change of the no-fly zone has affected the path; if the change of the no-fly zone has affected the current path, the first and last two points on the path that are not affected and are close to the no-fly zone are selected as a new starting point and an end point from the current position; for the new starting point and the end point, rasterization is performed in the current local map, and the same path planning algorithm is used to search in the new smaller raster map to obtain a new local path; and finally, the local path is spliced from part of the original path and the new local path.

[0046] The present application has the following beneficial effects:

[0047] 1. Task planning efficiency: the method realizes automatic allocation of the unmanned plane task through map initialization and analysis of the unmanned plane and the attack target, considers the influence of the kinematics of the unmanned plane and dynamic obstacles on the path, and improves the task execution efficiency.

[0048] 2. Precision and efficiency of multi-angle cooperative attack: The method considers the case of multiple UAVs attacking a target cooperatively, which can attack the same target from multiple angles, improving the task hit rate and effectiveness. Moreover, the method can automatically plan the path of each UAV, avoiding conflicts and collisions between UAVs, improving the safety of the task.

[0049] 3. Real-time of map information and dynamic obstacles: The method detects the no-fly area according to the current map information, and quickly plans the local path of the UAV. Therefore, the method has higher response speed, can timely process any changes, and improves the precision and efficiency of the task.

[0050] 4. Improved A* algorithm: The method proposes an improved A* algorithm, which can adapt to the grid size, ensure that the start and end points are on the endpoints of the grid, and reduce the inflection points, so that the UAV path can be planned faster, improving the response speed and accuracy of the task. The heuristic function is modified to better protect the safety of the UAV and effectively avoid interference between UAVs. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 A flow chart of a heterogeneous UAV cluster path planning method for multi-angle cooperative attack task is provided for the embodiments of the present application;

[0052] Figure 2 A battlefield situation illustration diagram is provided;

[0053] Figure 3 A local path planning diagram for the no-fly area is provided;

[0054] Figure 4 A running result diagram of the present application is provided. DETAILED DESCRIPTION

[0055] The present application will be described in detail below with reference to the accompanying drawings and examples.

[0056] The cooperative attack task is a very important link in the combat process. Due to the different types and performance of UAVs, task requirements and environmental constraints are different, resulting in different evaluations of each UAV for the same attack target. In the present application, all aerial UAVs are heterogeneous agents with different task execution capability ranges, and can select different payloads to perform tasks. The environment is an outdoor attack target area. The characteristics of the cooperative attack task include: each UAV needs to complete the cooperative task simultaneously, quickly and effectively, and the cooperative efficiency is maximized. The characteristics of the cooperative attack task need to be considered for task allocation, and finally a multi-angle cooperative attack on a certain target is realized.

[0057] In the cooperative attack task, the same target needs multiple UAVs to attack from different angles, so the UAV needs to be assigned a task before it performs the task. The present application proposes a method of initializing the map to determine the number of cooperative aircraft required for the attack task, and planning the path of all UAVs according to the map information. The task assignment of each UAV is evaluated according to the task requirements and its flight capability. After the task assignment, the improved A* algorithm is used to plan the global path of each UAV, considering the constraints such as turning radius and flight speed, and using Dubins path planning algorithm to convert it into a more suitable path in kinematic characteristics. At the same time, the present application also considers the influence of the no-fly zone, and when the no-fly zone changes, the affected path is planned locally to ensure that the flight path of the UAV can avoid the no-fly zone during the no-fly period.

[0058] As shown in Figure 1 , a heterogeneous UAV cluster path planning method for multi-angle cooperative attack task has the following specific steps:

[0059] Step 1: Information initialization, mainly including: UAV information and payload information initialization, terrain initialization, target initialization, enemy radar initialization.

[0060] Among them, the UAV information includes: UAV number, type, latitude and longitude, heading angle, minimum speed and maximum speed. The type mainly includes fixed-wing UAV and rotary-wing UAV. The payload information includes: attack payload number, attack method, default launch distance and number of ammunition.

[0061] The terrain information initialization includes: generating a series of obstacles, divided into fixed obstacles and obstacles with time window, the fixed obstacles save the latitude and longitude information of each vertex of the obstacle, and the obstacles with time window mainly refer to the no-fly zone, in addition to the geographical position information of each vertex, the start and end time of the existence of the obstacle is also added.

[0062] The target information initialization includes: generating a series of targets, the target information contains latitude and longitude geographical position information, the number of aircraft required for cooperative attack and the attack direction angle. The enemy radar initialization includes: initializing the latitude and longitude information of the enemy radar and the detection radius of the radar.

[0063] After initialization, the whole battlefield situation is shown in Figure 2 , which shows the attack target, irregular obstacle, enemy radar, our UAV and no-fly zone.

[0064] Step 2: According to the information after initialization, all UAVs are assigned a task to determine the target point of each UAV and the payload used, realizing multi-angle cooperative attack on multiple attack targets.

[0065] The greedy algorithm is used in the task allocation process, and the allocation process is as follows:

[0066] For one of the targets, the number of unmanned aerial vehicles that need to be allocated can be determined according to the number of aircrafts that need to be cooperatively attacked. For each task, the unmatched unmanned aerial vehicles and their loads are traversed, the evaluation value is calculated, the unmanned aerial vehicle-load pair with the lowest evaluation value is selected, and the current task is matched, then the attack point position is calculated according to the default attack distance of the target point information and the load information, and the following task chain is formed:

[0067] Target point-unmanned aerial vehicle-load-attack point-attack direction

[0068] The above process is repeated until the task allocation for all target points is completed.

[0069] The evaluation function involved in the task allocation is as follows:

[0070] Value = a · (dis-dis atack ) + b · (|heading uas -heading| + |heaeding attack -heading|)

[0071] Wherein a and b are the evaluation coefficients of distance and angle respectively, dis is the Euclidean distance of the unmanned aerial vehicle from the target point, dis atack is the default launch distance of the attack load, heading uas is the initial orientation angle of the unmanned aerial vehicle, heading attack is the attack direction angle of the attack target, and heading is the direction angle of the unmanned aerial vehicle pointing to the target point, wherein all distance variables are in meters and angle variables are in radians. In the calculation process, a is 0.001 and b is 57.296. The evaluation function takes into account the influence of the orientation angle, which can avoid large-angle deflection of the fixed-wing unmanned aerial vehicle when performing the task.

[0072] Step 3: Use the improved A* algorithm for path planning to get a path that can avoid obstacles, then consider the kinematics of the unmanned aerial vehicle, convert the existing path to a Dubins path, and finally achieve cooperative attack on the target with the optimal path.

[0073] First, the position of the unmanned aerial vehicle in the task chain is set as the starting point, the attack point is set as the end point, the map is rasterized, and the operation process is as follows. The starting point and the end point are the diagonal endpoints of the raster map, the raster map is set as a 40*40 rectangular grid, and then 20 grids are expanded outward on this basis, and finally an 80*80 rectangular grid map is obtained, and the starting point and the end point are just on the grid point. The grid size of the application can be automatically adjusted to ensure that the starting point and the target point of the path planning are on the end point of the grid, avoiding displacement of the starting point and the end point, and improving the accuracy of the path planning. When the target distance is far, the grid will also expand accordingly, which to some extent controls the search time of the A* algorithm.

[0074] Then the A* algorithm is used to find the optimal path in the raster map. It should be noted that only when the connecting line between the next grid point and the current grid point does not pass through the obstacle, the next grid point can be put into the open list. The heuristic function in the running process of the A* algorithm is designed as follows:

[0075] f(n)=g(n)+h(n)

[0076]

[0077] f(n) is the evaluation value of the priority of each grid point, g(n) is the cost of the grid point distance from the starting point, h(n) is the heuristic function of the A* algorithm, wherein dis represents the Manhattan distance of the grid point n from the end point, and the last three items are penalty items introduced for obstacles, enemy radars and other paths. The penalty item for the obstacle is:

[0078]

[0079]

[0080] wherein represents the Euclidean distance between the grid point n and the i-th obstacle vertex, R min represents the minimum hovering radius of the current unmanned aerial vehicle. It can be seen that when the distance from the obstacle vertex is more than 2R min +100, the value of the penalty item is 0, and when the distance is less than the threshold, the penalty item is introduced, and the closer the distance from the obstacle, the higher the penalty value. The penalty item can make the path obtained by the search maintain a certain safety distance from the obstacle.

[0081] The penalty item for the enemy radar is:

[0082]

[0083]

[0084] wherein disn,j represents the Euclidean distance between grid point n and the jth enemy radar, disj represents the detection radius of the jth enemy radar, it can be seen that when the distance from the enemy radar is greater than the threshold, the penalty term is 0, when the distance is less than the threshold, the penalty term is introduced, and the closer the distance from the radar, the higher the penalty value. The penalty term can effectively avoid the path into the detection area of the enemy radar.

[0085] The penalty term for other paths is:

[0086]

[0087]

[0088] wherein disn,k represents the Euclidean distance between grid point n and the kth other path point, dis' represents the Euclidean distance between the grid point and the end point, it can be seen that when the distance from the other path point is greater than the threshold, the penalty term is 0, when the distance is less than the threshold, the penalty term is introduced, and the closer the distance from the other path point, the higher the penalty value. The penalty term can avoid the intersection between two paths, when the distance from the end point is small, because the path is dense at this time, considering other paths may cause the phenomenon of oscillation, so dis' is introduced, when the distance is less than 2R min +200, the influence of other paths is not considered.

[0089] The above scheme modifies the heuristic function, taking into account the kinematic characteristics of the UAV, so that the UAV can avoid obstacles and enemy radars within a safe range, and avoid mutual collision between paths. It can better protect the safety of the UAV, and effectively avoid the interference between UAVs.

[0090] At this point, a series of path points can be obtained, but at this time the turning points of the path points are too many, if directly used for Dubins path calculation, a large amount of redundant path will be generated. Therefore, the path is considered to be smoothed, the inflection points with too close distance are deleted, and a new path with relatively smooth is obtained; the path is optimized, the calculation amount of Dubins path is reduced, and the calculation efficiency is improved. On the basis of the new path, the initial orientation, the final orientation, the minimum hovering radius and the flight speed are determined to obtain the Dubins path.

[0091] Step 4: The UAV starts to perform the task according to the respective Dubins path.

[0092] Step 5: In the process of executing the Dubins path by the UAV, it is detected whether the no-fly zone changes, when the no-fly zone changes (appears / disappears), the path affected is locally planned, and the constraints such as the turning radius of the UAV and the flight speed also need to be considered.

[0093] When the no-fly zone changes, firstly check whether the respective path is affected by the no-fly zone, if affected, select the first and last two points on the path which are not affected and close to the no-fly zone as the starting point and the ending point of the local path planning, and perform the local path planning in the same way, the only difference is that the calculation amount is reduced in the process of rasterizing the map, the starting point and the ending point are the opposite endpoints of the raster map, the raster map is set as a 20*20 rectangular grid, and then 10 grids are expanded outward on this basis, finally a 40*40 rectangular grid map is obtained. In this way, the path search time can be greatly shortened, and rapid local path planning is realized.

[0094] Figure 3 The influence of the no-fly zone change on the path is shown, when the no-fly zone exists, the unmanned aerial vehicle selects to bypass the no-fly zone, in the bypassing process, the no-fly zone is cancelled, and after local path planning, the unmanned aerial vehicle can pass through the area again.

[0095] Step 6: all unmanned aerial vehicles complete the attack task, the running result graph is as shown in Figure 4 , and the task ends.

[0096] The present application is aimed at the demand of multi-target cooperative attack task, the improved A* algorithm is used to realize the global path planning of the unmanned aerial vehicle, and then the path is converted into a Dubins path and multi-angle cooperative attack on multiple attack targets is realized. The method comprehensively considers, optimizes and integrates multiple factors, realizes the efficiency, safety of the path planning of the unmanned aerial vehicle and the demand of the cooperative attack task, and has wide application prospect.

[0097] The present application is aimed at the demand of multi-target cooperative attack task, the improved A* algorithm is used to realize the global path planning of the unmanned aerial vehicle, and then the path is converted into a Dubins path. Then, multi-angle cooperative attack on multiple attack targets is realized through static planning and dynamic planning. The local path planning for the no-fly zone is considered. When the no-fly zone changes, firstly check whether the respective path is affected by the no-fly zone. If the path of the unmanned aerial vehicle is affected, the method selects the first and last two points on the path which are not affected and close to the no-fly zone as the starting point and the ending point of the local path planning. Then, the local path planning is performed on the sub-region, and rapid local path planning is realized. In terms of realizing the efficiency, safety of the path planning of the unmanned aerial vehicle and the demand of the cooperative attack task, the method comprehensively considers, optimizes and integrates multiple factors, and has wide application prospect.

[0098] To sum up, the above is only a preferred embodiment of the present application, and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A heterogeneous UAV swarm path planning method for multi-angle cooperative strike missions, characterized in that, Includes the following steps: Step 1: Perform information initialization, including: UAV and payload information initialization, terrain initialization, target initialization, and enemy radar initialization; Step 2: Based on the initialized information, assign tasks to coordinate multi-angle strikes against multiple targets; Step 3: Use the improved A* algorithm for path planning, and then consider the kinematic characteristics of the UAV to finally convert the path into a Dubins path; First, the drone position in the mission chain is set as the starting point and the strike point as the ending point. The map is then rasterized, with the starting point and the ending point as the diagonal endpoints of the raster map. Then, the A* algorithm is used to find the optimal path in the raster map. The heuristic function h(n) used in the improved A* algorithm is designed as follows: f(n) = g(n) + h(n) Where f(n) is the priority evaluation value of each grid point n, g(n) is the cost of grid point n from the starting point, and dis represents the Manhattan distance of grid point n from the ending point. The penalty terms for obstacles are: in, R represents the Euclidean distance between grid point n and the i-th obstacle vertex. min This indicates the current minimum hovering radius of the drone; The penalties for enemy radar include: in, This represents the Euclidean distance between grid point n and the j-th enemy radar. This represents the detection radius of the j-th enemy radar; The penalties for other paths are: in, The distance between grid point n and the kth other planned path point is represented by the Euclidean distance, and dis' represents the Euclidean distance between grid point n and the destination. Step 4: The drones begin executing their missions according to their respective Dubins paths; Step 5: When the no-fly zone changes, check if the flight path is affected. If it is, perform local flight path planning and adjust the flight path accordingly. Step 6: All drones have completed their strike missions, and the mission is over.

2. The method as described in claim 1, characterized in that, In the improved A* algorithm, the grid size is adjusted as needed to ensure that the starting point and target point of the path planning are both at the endpoints of the grid.

3. The method as described in claim 2, characterized in that, Set the grid map to a 40*40 rectangular grid, and then expand it outward by 20 grids to obtain an 80*80 rectangular grid map, with the start and end points exactly on the grid points.

4. The method as described in claim 1, characterized in that, The improved A* algorithm smooths the generated path and removes redundant inflection points when calculating the Dubins path.

5. The method as described in claim 1, characterized in that: The information includes UAV and payload information, terrain information, target information, and enemy radar information; The drone information includes drone number, type, latitude and longitude, heading angle, minimum speed, and maximum speed; The types of drones include fixed-wing drones and rotary-wing drones; The payload information includes attack payload number, attack method, default launch distance, and number of missiles; The payload strike methods include guided and unguided methods; The terrain information includes the latitude, longitude, and altitude information of each vertex of the fixed obstacle, the no-fly zone no-fly time range with time windows, and the latitude, longitude, and altitude information of each vertex of the no-fly zone; The target information includes the latitude, longitude, and altitude of the target point, the number of aircraft that need to be coordinated for the strike, and the direction and angle of the strike. The enemy radar information includes the enemy radar's latitude, longitude, altitude, and detection radius.

6. The method as described in claim 1, characterized in that, The specific process of task allocation is as follows: First, determine the number of drones that need to be coordinated to attack each target, and calculate which drone is best matched with the current task based on the evaluation function; then, calculate the attack point position and attack direction that each drone should attack based on the default attack distance of the target point information and payload information; repeat this process until task allocation is completed for all target points.

7. The method as described in claim 6, characterized in that, The evaluation function used in the task allocation process is as follows: Value=α·(dis-dis atack )+β·(|heading uas -heading|+|heading attack -heading|) Where α and β are the evaluation coefficients for distance and angle, respectively, and dis is the Euclidean distance between the UAV and the target point. atack The default launch range for the attack payload, heading uas The initial heading angle of the drone. attack The heading is the angle at which the drone points towards the target.

8. The method as described in claim 7, characterized in that, In the task allocation process, α is set to 0.001 and β is set to 57.

296.

9. The method as described in claim 1, characterized in that, The local path planning process includes: first, calculating an initial path using the improved A* algorithm based on the current and target locations; then, continuously checking whether changes in no-fly zones affect the path during local path planning; if changes in no-fly zones affect the current path, selecting two unaffected points on the path that are close to the no-fly zone as the new start and end points, starting from the current location; for the new start and end points, rasterizing the current local map, and searching using the same path planning algorithm on the new, smaller raster map to obtain a new local path; the final local path is formed by combining a part of the original path and the new local path.

Images