Path planning method, device, equipment and medium for drone swarm
By using the path map to plan global paths and determine real-time key points in the drone group, and optimizing paths with environmental obstacle information, the problem of low path selection efficiency during flight of the drone group is solved, and obstacle avoidance of the drone group is improved Capacity and overall efficiency.
Patent Information
- Application Number
- CN202211062597.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-08-31
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2042-08-31
AI Technical Summary
The existing drone swarm path planning algorithm cannot fully consider the relationship between the global path of the drone swarm and the flight path of a single drone, resulting in some drones being unable to choose reasonable paths in time during flight, which reduces efficiency.
Global path planning is carried out based on the path map within the moving range of the drone group, and the key point coordinates of the drone group are determined in real time. Local optimal trajectories are screened based on environmental obstacle information, and local optimization of the global path is carried out to adjust the flight path of the drone group. .
By adjusting the path of the drone cluster in real time, the obstacle avoidance and overall efficiency of the aircraft cluster when encountering obstacles are improved.
Smart Images

Figure CN115268502B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of unmanned aerial vehicle (UAV) technology, and in particular to a path planning method, device, equipment and medium for a swarm of unmanned aerial vehicles (UAVs). Background Art
[0002] Drone swarm is a new drone working mode proposed in recent years. It is inspired by the swarm behavior of relatively low-intelligence creatures such as bee swarms, fish schools, and bird flocks during migration, patrolling, or avoiding enemies. By taking advantage of the swarm, it can present overall behavior based on simple rules and coordination, and improve the efficiency of completing inspections and other tasks with the advantage of numbers. It is widely used in remote sensing detection, remote inspections and other mission scenarios.
[0003] In the working mode of drone swarm, each drone can be regarded as an independent positioning node. The commonly used drone swarm path planning algorithm cannot fully consider the relationship between the global path of the drone swarm and the flight path of a single drone. In addition, since the drone swarm is easily affected by environmental interference factors during flight, some drones may not be able to choose a reasonable path in time when avoiding dense obstacles, resulting in reduced swarm efficiency. Summary of the invention
[0004] Based on this, it is necessary to provide a path planning method, device, equipment and medium for a swarm of drones to solve the above problems.
[0005] A path planning method for a drone swarm, the method comprising:
[0006] Perform global path planning based on the path map within the movement range of the drone swarm to obtain the global path of the drone swarm;
[0007] After the drone swarm starts to move based on the global path, the coordinates of the key points of the drone swarm at the current moment are determined in real time based on the positional relationship between the drones in the drone swarm;
[0008] According to the coordinates of key points of the drone swarm at the current moment, multiple groups of local predicted trajectories are determined, and local optimal trajectories are selected from the multiple groups of local predicted trajectories according to the environmental obstacle information perceived by the drone swarm. The global path is locally optimized based on the local optimal trajectory, and the drone swarm is driven to continue moving based on the optimized global path.
[0009] In one embodiment, after the drone group starts to move based on the global path, determining the key point coordinates of the drone group at the current moment based on the position relationship between each drone in the drone group in real time includes:
[0010] At the current moment, a first coordinate of a target UAV and a second coordinate of a neighboring UAV in a two-dimensional space are obtained, and a positioning diffusion degree between the target UAV and the neighboring UAV is calculated according to the first coordinate and the second coordinate; wherein the target UAV is any UAV in the UAV group, the neighboring UAV is any UAV in the UAV group except the target UAV, and the positioning diffusion degree is used to measure the distance between the target UAV and the neighboring UAV;
[0011] Determine a diffusion matrix of the drone group based on the positioning diffusion between any two drones in the drone group, and calculate the proximity of the target drone based on the diffusion matrix; wherein the diffusion matrix is used to indicate the size of the distribution range of the drone group, and the proximity of the target drone indicates the degree of proximity between the first coordinate and all the second coordinates;
[0012] Calculate the positioning mean and positioning error of the target UAV according to all approximations of the target UAV during its movement;
[0013] The individual weight of the target UAV relative to the UAV swarm is calculated according to the positioning mean and positioning error of the target UAV, so as to obtain the individual weights of all UAVs relative to the UAV swarm;
[0014] The key point coordinates of the drone swarm at the current moment are calculated according to the individual weights of all drones relative to the drone swarm, and the first coordinates and the second coordinates.
[0015] In one embodiment, the environmental obstacle information includes the distance between the local predicted trajectory and the obstacle, and the determining of multiple groups of local predicted trajectories according to the key point coordinates of the drone group at the current moment, and selecting the local optimal trajectory from the multiple groups of local predicted trajectories according to the environmental obstacle information perceived by the drone group, includes:
[0016] The coordinates of the key points of the drone group at the current moment are used as the local starting point, a preset dynamic window is set with the local starting point as the starting reference point, and a local end point is determined within the preset dynamic window according to the global path;
[0017] Simulating multiple groups of local predicted trajectories of the drone swarm between the local starting point and the local end point according to the current speed of the drone swarm;
[0018] The multiple groups of local predicted trajectories are scored according to the distances of the multiple groups of local predicted trajectories from obstacles and a preset local path evaluation function, and a group of local predicted trajectories with the highest score is selected as the local optimal trajectory; wherein the local path evaluation function includes a static obstacle avoidance path scoring function and a dynamic obstacle avoidance path scoring function, the static obstacle avoidance path scoring function is the product of a first dynamic weight and a first distance, the dynamic obstacle avoidance path scoring function is the product of a second dynamic weight and a second distance, the first distance is the distance between a group of local predicted trajectories and a static obstacle, and the second distance is the distance between a group of local predicted trajectories and a dynamic obstacle.
[0019] In one embodiment, after simulating multiple groups of local predicted trajectories of the drone group between the local starting point and the local end point according to the current speed of the drone group, the method further includes:
[0020] At the current moment, the first dynamic weight is set for the ratio of the number of static obstacles to all obstacles, and the second dynamic weight is set for the ratio of the number of dynamic obstacles to all obstacles.
[0021] In one embodiment, performing global path planning based on a path map within the moving range of the drone swarm to obtain a global path of the drone swarm includes:
[0022] Connect the global starting point and the global end point set in the path map with a random route to generate the initial path of the drone swarm, and use the initial path of the drone swarm as the current path of the drone swarm;
[0023] The current fitness value of the drone swarm is calculated based on a preset fitness function and the current path of the drone swarm, the current fitness value of the target drone is used as the current individual optimal value of the target drone, and the current minimum fitness value is used as the current group optimal value of the drone swarm; wherein the target drone is any drone in the drone swarm, and the current minimum fitness value is the minimum value among the current fitness values of the drone swarm;
[0024] Update the current path of the drone swarm according to the current individual optimal value of the target drone and the current group optimal value of the drone swarm to obtain an updated path of the drone swarm;
[0025] Determine whether the number of current update paths is equal to the preset maximum number of iterations;
[0026] If the number of the current update path is less than the preset maximum number of iterations, the current fitness value of the drone swarm is calculated based on the preset fitness function and the update path of the drone swarm. If the updated current fitness value of the target drone is less than the current individual optimal value of the target drone, the current individual optimal value of the target drone is updated with the updated current fitness value of the target drone. If the updated current minimum fitness value is less than the current group optimal value of the drone swarm, the current group optimal value of the drone swarm is updated with the updated current minimum fitness value.
[0027] Taking the updated path of the drone swarm as the current path of the drone swarm, returning to execute the step of updating the current path of the drone swarm according to the current individual optimal value of the target drone and the current group optimal value of the drone swarm to obtain the updated path of the drone swarm and subsequent steps;
[0028] If the number of times the current path is updated is equal to the preset maximum number of iterations, the updated path of the drone swarm currently obtained is output as the global path.
[0029] In one embodiment, updating the current path of the drone swarm according to the current individual optimal value of the target drone and the current group optimal value of the drone swarm includes:
[0030] The drone swarm is divided into discoverers and followers according to the current individual optimal values of all drone swarms; wherein the number ratio between the discoverers and the followers is a preset ratio, and the current fitness values of the discoverers are all less than the current fitness values of the followers;
[0031] Update the current path of the discoverer based on the security value;
[0032] Update the current path of the follower according to the path of the drone corresponding to the current optimal value of the group and the path of the drone corresponding to the current worst value of the group; wherein the current worst value of the group is the maximum value among the current fitness values of the drone group;
[0033] A preset number of drones are randomly selected from the drone group as early warning detectors, and the current path of the early warning detector is updated according to the path of the drone corresponding to the current optimal value of the group and the path of the drone corresponding to the current worst value of the group.
[0034] In one embodiment, the method further comprises:
[0035] According to the global path, a fuzzy obstacle avoidance rule and a fuzzy guidance rule of the drone swarm are constructed; wherein the input of the fuzzy obstacle avoidance rule is the distance between the drone and the obstacle in each direction, and the output of the fuzzy obstacle avoidance rule is a preset fuzzy distance; the input of the fuzzy guidance rule is the angle and distance between the drone and the global end point, and the output of the fuzzy guidance rule is a preset fuzzy direction.
[0036] A path planning device for a drone group, the device comprising:
[0037] The global planning module is used to perform global path planning based on the path map within the movement range of the drone swarm to obtain the global path of the drone swarm;
[0038] A real-time key point acquisition module is used to determine the key point coordinates of the drone group at the current moment based on the positional relationship between each drone in the drone group in real time after the drone group starts to move based on the global path;
[0039] The local optimization module is used to determine multiple groups of local predicted trajectories according to the key point coordinates of the drone swarm at the current moment, and select the local optimal trajectory from the multiple groups of local predicted trajectories according to the environmental obstacle information perceived by the drone swarm, locally optimize the global path based on the local optimal trajectory, and drive the drone swarm to continue moving based on the optimized global path.
[0040] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the steps of the above-mentioned path planning method for a drone swarm.
[0041] A path planning device for a drone swarm includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above-mentioned path planning method for the drone swarm.
[0042] The present invention provides a path planning method, device, equipment and medium for a drone swarm. First, a global path planning is performed based on a path map within the moving range of the drone swarm to obtain a global path of the drone swarm. After the drone swarm starts to move based on the global path, the key point coordinates of the drone swarm at the current moment are determined in real time based on the positional relationship between each drone in the drone swarm, so as to locate the drone swarm by integrating the position of each drone at the current moment, providing a basis for the subsequent track planning of the drone swarm. Finally, multiple groups of local predicted trajectories are determined based on the key point coordinates of the drone swarm at the current moment, and local optimal trajectories are selected from multiple groups of local predicted trajectories based on the environmental obstacle information perceived by the drone swarm, and the global path is locally optimized based on the local optimal trajectory, and the drone swarm is driven to continue moving based on the optimized global path. In this way, the global path of the drone swarm path can be adjusted in real time based on the obstacles encountered locally, so that each drone can avoid obstacles that appear in real time and improve the overall efficiency of the swarm. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0044] in:
[0045] Figure 1 A schematic diagram of a flow chart of a path planning method for a drone swarm in one embodiment;
[0046] Figure 2 A schematic diagram of a global path planning process in a specific embodiment;
[0047] Figure 3 A schematic diagram of a process for updating a current path of a drone swarm in a specific embodiment;
[0048] Figure 4 A schematic diagram of a process for determining key point coordinates in a specific embodiment;
[0049] Figure 5 It is a schematic diagram of a process of local path planning in a specific embodiment;
[0050] Figure 6 It is a structural schematic diagram of a path planning device for a drone swarm in one embodiment;
[0051] Figure 7 A structural block diagram of a path planning device for a drone swarm in one embodiment. DETAILED DESCRIPTION
[0052] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0053] like Figure 1 As shown, Figure 1 FIG. 1 is a flow chart of a method for planning a path for a drone swarm in one embodiment. The method for planning a path for a drone swarm in this embodiment includes the following steps:
[0054] Step 102, performing global path planning based on the path map within the movement range of the drone swarm to obtain the global path of the drone swarm.
[0055] That is, based on the mastered path map, the process of planning in advance how the drone swarm moves from the global starting point to the global end point.
[0056] In a specific embodiment, Figure 2 As shown, global path planning is performed through the following specific steps, including:
[0057] Step 1021, connect the global starting point and the global end point set in the path map with a random route to generate an initial path for the drone swarm, and use the initial path of the drone swarm as the current path of the drone swarm.
[0058] Among them, the positions of various static obstacles are marked in the path map, and the global starting point and global end point in the path map are manually set and can be set according to the actual needs of the user. It is understandable that the initial path generated here must not conflict with static obstacles, and the initial path will be continuously optimized in subsequent steps until it meets the actual path planning needs.
[0059] Step 1022, calculate the current fitness value of the drone swarm based on the preset fitness function and the current path of the drone swarm, take the current fitness of the target drone as the current individual optimal value of the target drone, and take the current minimum fitness value as the current group optimal value of the drone swarm.
[0060] The target drone is any drone in the drone group, that is, the same operation is performed on all drones. The current minimum fitness value is the minimum value among the current fitness values of the drone group.
[0061] Specifically, assuming that the drone swarm includes N drones, the current path of the drone swarm can be expressed by a matrix:
[0062] X=[x1,x2,...,x N ] T ,x i =[x i,1 ,x i,2 ,...,x i,d ],i=1,2,...,N (1)
[0064] In the above formula, x i represents the current path of the i-th drone in the drone swarm, and d is the dimension of the problem to be optimized, that is, the number of variables to be optimized, such as path length, smoothness, etc.
[0065] Furthermore, a fitness function can be pre-set based on these optimization problems. Here, the fitness function is generally defined as follows: the smaller the calculated fitness value, the better the problem is optimized. Taking the path length as an example, it can be set that the smaller the fitness value calculated by the fitness function, the shorter the path length is. After substituting the current path of the drone group into the preset fitness function, the current fitness value of the drone group can be calculated, which can be expressed as:
[0066]
[0067] in, is the current fitness value of the i-th drone in the drone swarm when the fitness value is calculated for the first time.
[0068] Next, set the current individual optimal value and the current group optimal value. For example, if the drone numbered 1 is regarded as the target drone, then the current As the current individual optimal value of the drone numbered 1. If the drone numbered 2 is regarded as the target drone, then the current As the current individual optimal value of the drone numbered 2. The same applies to the other drones, so I will not repeat them here. yes The minimum value in As the current group optimal value of the drone swarm. It can be understood that the current individual optimal value and the current group optimal value here will be continuously updated as the subsequent steps proceed.
[0069] Step 1023, updating the current path of the drone swarm according to the current individual optimal value of the target drone and the current group optimal value of the drone swarm to obtain an updated path of the drone swarm.
[0070] In a specific embodiment, Figure 3 As shown, the current path of the drone swarm can be updated through the following specific steps:
[0071] Step 1023a: Divide the UAV swarm into discoverers and followers according to the current individual optima of all UAV swarms.
[0072] Among them, the quantity ratio between the discoverers and the followers is a preset ratio, and the current fitness values of the discoverers are all smaller than those of the followers. For example, the quantity ratio between the discoverers and the followers is set to 2:8.
[0073] It can be understood that here, the individuals that can find better paths (UAVs with smaller current individual optima) are used as discoverers, and the individuals that cannot find better paths (UAVs with larger current individual optima) are used as followers. In the subsequent specific steps, the discoverers will provide the flight area and direction guidance for the followers, so as to achieve the update effect of "driving the inferior with the superior" and realize the overall path update of the UAV swarm.
[0074] Step 1023b: Update the current path of the discoverers based on the safety value.
[0075] Specifically, the path update process of the discoverers can be expressed as:
[0076]
[0077] In the above formula, t is the current iteration number; Tmax is the preset maximum iteration number; is the current path of the i-th UAV in the j-th dimension at the iteration number t. In this embodiment, it is limited to a two-dimensional plane; R2 is between [0,1], representing the warning value; ST is between [0.5,1], representing the safety value; α is a random number belonging to (0,1); Q is a normal distribution random number between [0,1]; L is a 1xd identity matrix.
[0078] When R2 < ST, this means it is safe at this time, and the discoverers can perform extensive search operations. If R2 ≥ ST, this indicates danger at this time, and the discoverers issue an alarm at this time, and all discoverers need to quickly fly to other safe places to forage.
[0079] Step 1023c: Update the current path of the followers according to the paths of the UAVs corresponding to the current global optimum and the paths of the UAVs corresponding to the current global worst value.
[0080] Among them, the current global worst value is the maximum value among the current fitness values of the UAV swarm.
[0081] Specifically, the path update process of the followers can be expressed as:
[0082]
[0083] In the above formula, is the path of the UAV corresponding to the worst value of the current group when the number of iterations is t; is the path of the drone corresponding to the current optimal value of the group when the number of iterations is t+1; + A is a matrix whose elements are randomly assigned 1 or -1, where A + =A T (AA T ) -1 .
[0084] when This indicates that the i-th drone joiner with a lower fitness value did not obtain a better path and needs to fly to other places to find food in order to obtain a better path.
[0085] Step 1023d, randomly select a preset number of drones from the drone group as early warning devices, and update the current path of the early warning device according to the path of the drone corresponding to the current optimal value of the group and the path of the drone corresponding to the current worst value of the group.
[0086] In this embodiment, k drones (for example, 10% to 20% of the total number of drones) are randomly selected from the drone population for early warning reconnaissance. Specifically, the path update process of the early warning person can be expressed as:
[0087]
[0088] In the above formula, f g is the optimal value of the current group; β is the step size control parameter; K is a random number between [-1,1], which obeys a uniform distribution and is used to control the direction of the UAV movement; f i is the current fitness value of UAV i; ε is the minimum constant; f w is the worst value of the current group. i >f g When f i =f g , it indicates that the drone is located in the center of the population and can lead followers to plan paths.
[0089] Step 1024, determine whether the number of times the current path is updated is equal to the preset maximum number of iterations. If the number of times the current path is updated is less than the preset maximum number of iterations, execute step 1025; if the number of times the current path is updated is equal to the preset maximum number of iterations, execute step 1031.
[0090] Step 1025, calculating the current fitness value of the drone swarm based on the preset fitness function and the update path of the drone swarm.
[0091] That is, the updated path of the drone swarm is substituted into the preset fitness function to update the current fitness value of the drone swarm, which can be expressed as:
[0092]
[0093] in, is the current fitness value of the i-th drone in the drone group when the fitness value is calculated for the a-th time. For example, for the first update, which is equivalent to the second calculation of the current fitness value, it is expressed as
[0094] Step 1026, determine whether the updated current fitness value of the target drone is less than the current individual optimal value of the target drone. If the updated current fitness value of the target drone is less than the current individual optimal value of the target drone, execute step 1027 to update the current individual optimal value of the target drone with the updated current fitness value of the target drone.
[0095] For example, if the drone numbered 1 is regarded as the target drone, after the first update, it is necessary to determine Is it less than like Less than Then As the current individual optimal value of the drone numbered 1. Otherwise, no replacement is made and As the current individual optimal value of the drone numbered 1. Of course, the same is true for other drones, so I will not go into details.
[0096] Step 1028, determine whether the updated current minimum fitness value is less than the current group optimal value of the drone group. If the updated current minimum fitness value is less than the current group optimal value of the drone group, execute step 1029 to update the current group optimal value of the drone group with the updated current minimum fitness value.
[0097] For example, for a drone group, if before the update, is the current optimal value of the drone swarm. After the first update, if is the current minimum fitness value after update. Is it less than like Less than Then As the current optimal value of the drone group. Otherwise, no replacement will be made. As the current optimal value of the drone swarm.
[0098] Step 1030, taking the updated path of the drone swarm as the current path of the drone swarm, and returning to execute step 1023 and subsequent steps.
[0099] Step 1031, output the updated path of the currently acquired drone swarm as the global path.
[0100] Step 104, after the drone swarm starts to move based on the global path, the coordinates of the key points of the drone swarm at the current moment are determined in real time based on the positional relationship between the drones in the drone swarm.
[0101] In the actual movement of the drone swarm, each drone can be regarded as an independent node, and due to the influence of factors such as the environment and electromagnetic interference, the overall positioning of the drone swarm often deviates. To solve this problem, the coordinates of the key points of the drone swarm at the current moment are determined in real time based on the position relationship between each drone in the drone swarm, so as to locate the drone swarm by combining the position of each drone at the current moment, providing a basis for the subsequent trajectory planning of the drone swarm.
[0102] In a specific embodiment, Figure 4 As shown, the specific steps to determine the coordinates of the key points are as follows:
[0103] Step 1042, at the current moment, obtain the first coordinates of the target UAV and the second coordinates of the neighboring UAVs in the two-dimensional space, and calculate the positioning diffusion between the target UAV and the neighboring UAVs based on the first coordinates and the second coordinates.
[0104] The target drone is any drone in the drone group, and the neighboring drone is any drone in the drone group except the target drone.
[0105] For example, for the i-th target drone node, its coordinates at time t in two-dimensional space can be expressed as (x i (t),y i (t)). Then the positioning diffusion between the target UAV i and the neighboring UAV j can be expressed as:
[0106]
[0107] Among them, the positioning diffusion is used to measure the distance between the target UAV and the neighboring UAVs.
[0108] Step 1044, determining a diffusion matrix of the drone group based on the positioning diffusion between any two drones in the drone group, and calculating the proximity of the target drone based on the diffusion matrix.
[0109] Among them, the diffusion matrix is used to indicate the distribution range of the drone group, and the diffusion matrix is expressed as:
[0110]
[0111] The approximation of the target drone indicates the degree of approximation between the first coordinate and all the second coordinates. At time t, the approximation between drone i and other drone coordinates is expressed as:
[0112]
[0113] Step 1046, calculating the positioning mean and positioning error of the target UAV according to all approximations of the target UAV during its movement.
[0114] Specifically, locate the mean and positioning error Respectively expressed as:
[0115]
[0116]
[0117] In the above formula, T is the total number of moments.
[0118] Step 1048, calculating the individual weight of the target UAV relative to the UAV swarm based on the positioning mean and positioning error of the target UAV, so as to obtain the individual weights of all UAVs relative to the UAV swarm.
[0119] Specifically, the individual weight of the target UAV relative to the UAV swarm is expressed as:
[0120]
[0121] Of course, after all drones in the drone swarm execute steps 1042-1048, the individual weights of all drones relative to the drone swarm can be obtained.
[0122] Step 1050, calculating the key point coordinates of the drone group at the current moment according to the individual weights of all drones relative to the drone group, and the first coordinate and the second coordinate.
[0123] Specifically, the coordinates of the key points of the drone swarm at the current moment can be expressed as:
[0124]
[0125] Using this method, the coordinates of key points in the drone swarm can be better located by effectively fusing the positioning information of multiple nodes, thereby achieving real-time and accurate positioning of the drone swarm.
[0126] Step 106, determine multiple groups of local predicted trajectories based on the key point coordinates of the drone swarm at the current moment, select local optimal trajectories from the multiple groups of local predicted trajectories based on the environmental obstacle information perceived by the drone swarm, locally optimize the global path based on the local optimal trajectory, and drive the drone swarm to continue moving based on the optimized global path.
[0127] When the UAV performs inspection tasks, it is not only necessary to plan the global path first, but also to establish a local path planning method for local dynamic obstacles during the inspection process.
[0128] In a specific embodiment, Figure 5 As shown, the specific steps of local path planning include:
[0129] Step 1062, taking the key point coordinates of the drone group at the current moment as the local starting point, setting a preset dynamic window with the local starting point as the starting reference point, and determining the local end point within the preset dynamic window according to the global path.
[0130] The size and setting position of the preset dynamic window are preset. The local end point is located in the preset dynamic window. If the global path passes through the currently set preset dynamic window, the end point of the global path in the preset dynamic window is set as the local end point. If the global path does not pass through the currently set preset dynamic window, the point in the preset dynamic window closest to the global path is set as the local end point.
[0131] Step 1064 , simulating multiple groups of local predicted trajectories of the drone swarm between the local starting point and the local end point according to the current speed of the drone swarm.
[0132] Step 1066 , scoring the multiple groups of local predicted trajectories according to the distances of the multiple groups of local predicted trajectories from the obstacles and the preset local path evaluation function, and selecting the group of local predicted trajectories with the highest score as the local optimal trajectory.
[0133] In a specific embodiment, the local path evaluation function is expressed as:
[0134] G(v,ω)=β·dist s (v,ω)+τ·dist a (v,ω) (14)
[0135] In the above formula, the local path evaluation function includes the static obstacle avoidance path scoring function β·dist s (v,ω) and dynamic obstacle avoidance path scoring function τ·dist a (v,ω). The static obstacle avoidance path scoring function is the first dynamic weight β and the first distance dist sThe dynamic obstacle avoidance path scoring function is the product of the second dynamic weight τ and the second distance dist a The first distance is the distance between a set of local predicted trajectories and a static obstacle, and the second distance is the distance between a set of local predicted trajectories and a dynamic obstacle.
[0136] Compared with the traditional dynamic window optimization method, the local path evaluation function proposed in this specific embodiment can dynamically adjust the first dynamic weight β and the second dynamic weight τ based on the type of obstacle currently encountered. When the drone encounters a dynamic obstacle during the local path planning process, the obstacle avoidance capability of the dynamic obstacle can be improved by increasing the second dynamic weight τ. Conversely, when the drone does not encounter a dynamic obstacle during the local path planning process, the obstacle avoidance capability of the static obstacle can be improved by increasing the first dynamic weight β.
[0137] Specifically, the method of dynamically adjusting the first dynamic weight β and the second dynamic weight τ may include: at the current moment, setting the first dynamic weight as the ratio of the number of static obstacles to all obstacles, and setting the second dynamic weight as the ratio of the number of dynamic obstacles to all obstacles.
[0138] Furthermore, by substituting the distances of these multiple sets of local predicted trajectories from obstacles into formula (14), we can obtain the scores of multiple sets of local predicted trajectories, and finally select the set of local predicted trajectories with the highest scores as the local optimal trajectory. In this way, the interference caused by static obstacles and dynamic obstacles can be effectively reduced, and the accuracy and stability of obstacle avoidance can be improved.
[0139] Furthermore, fuzzy obstacle avoidance rules and fuzzy guidance rules for the drone group can also be constructed based on the global path. The fuzzy obstacle avoidance rules and fuzzy guidance rules are used to control the actual flight direction of the drone. Among them, the input of the fuzzy obstacle avoidance rule is the distance between the drone and the obstacle in each direction, including the obstacle distance in three directions (dl, d, dr); the output of the fuzzy obstacle avoidance rule is a preset fuzzy distance, including (near (N), middle (M), far (F)). The input of the fuzzy guidance rule is the angle and distance between the drone and the global end point, and the output of the fuzzy guidance rule is a preset fuzzy direction, including (left (N-), left middle (N), middle (Z), right middle (P), right (P+)). Based on this process, the input can be fuzzified into fuzzy data, fuzzy rules can be made using empirical theory, and fuzzy output can be obtained through certain calculations. In this way, the performance of the control system can be changed automatically by the environment in real time, and it has the characteristics of strong robustness.
[0140] The above-mentioned path planning method for the drone swarm first performs global path planning based on the path map within the moving range of the drone swarm to obtain the global path of the drone swarm. After the drone swarm starts to move based on the global path, the key point coordinates of the drone swarm at the current moment are determined in real time based on the position relationship between the drones in the drone swarm, so as to locate the drone swarm based on the position of each drone at the current moment, providing a basis for the subsequent trajectory planning of the drone swarm. Finally, multiple groups of local predicted trajectories are determined based on the key point coordinates of the drone swarm at the current moment, and the local optimal trajectory is selected from the multiple groups of local predicted trajectories based on the environmental obstacle information perceived by the drone swarm, and the global path is locally optimized based on the local optimal trajectory, and the drone swarm is driven to continue moving based on the optimized global path. In this way, the global path of the drone swarm path can be adjusted in real time based on the obstacles encountered locally, so that each drone can avoid obstacles that appear in real time and improve the overall efficiency of the swarm.
[0141] In one embodiment, Figure 6 As shown, a path planning device for a drone swarm is proposed, the device comprising:
[0142] A global planning module 602 is used to perform global path planning based on a path map within the movement range of the drone swarm to obtain a global path of the drone swarm;
[0143] A real-time key point acquisition module 604 is used to determine the key point coordinates of the drone group at the current moment based on the positional relationship between each drone in the drone group in real time after the drone group starts to move based on the global path;
[0144] The local optimization module 606 is used to determine multiple groups of local predicted trajectories based on the key point coordinates of the drone group at the current moment, and select the local optimal trajectory from the multiple groups of local predicted trajectories based on the environmental obstacle information perceived by the drone group, locally optimize the global path based on the local optimal trajectory, and drive the drone group to continue moving based on the optimized global path.
[0145] Figure 7 FIG. 1 shows an internal structure diagram of a path planning device for a drone swarm in one embodiment. Figure 7 As shown, the path planning device of the drone swarm includes a processor, a memory and a network interface connected via a system bus. Among them, the memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the path planning device of the drone swarm stores an operating system and may also store a computer program. When the computer program is executed by the processor, the processor can implement the path planning method of the drone swarm. The internal memory may also store a computer program. When the computer program is executed by the processor, the processor can execute the path planning method of the drone swarm. Those skilled in the art can understand that Figure 7 The structure shown in the figure is only a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the path planning device of the drone swarm to which the scheme of the present application is applied. The specific path planning device of the drone swarm may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0146] A path planning device for a drone swarm comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are implemented: global path planning is performed based on a path map within the movement range of the drone swarm to obtain a global path of the drone swarm; after the drone swarm starts to move based on the global path, the key point coordinates of the drone swarm at the current moment are determined in real time based on the positional relationship between each drone in the drone swarm; multiple groups of local predicted trajectories are determined based on the key point coordinates of the drone swarm at the current moment, and local optimal trajectories are selected from the multiple groups of local predicted trajectories based on environmental obstacle information perceived by the drone swarm, the global path is locally optimized based on the local optimal trajectory, and the drone swarm is driven to continue moving based on the optimized global path.
[0147] A computer-readable storage medium stores a computer program, which implements the following steps when executed by a processor: performing global path planning based on a path map within the movement range of a drone group to obtain a global path of the drone group; after the drone group starts to move based on the global path, determining the key point coordinates of the drone group at the current moment in real time based on the positional relationship between each drone in the drone group; determining multiple groups of local predicted trajectories based on the key point coordinates of the drone group at the current moment, and selecting local optimal trajectories from the multiple groups of local predicted trajectories based on environmental obstacle information perceived by the drone group, locally optimizing the global path based on the local optimal trajectory, and driving the drone group to continue moving based on the optimized global path.
[0148] It should be noted that the above-mentioned path planning method, device, equipment and computer-readable storage medium for a swarm of drones belong to a general inventive concept, and the contents of the embodiments of the path planning method, device, equipment and computer-readable storage medium for a swarm of drones are applicable to each other.
[0149] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0150] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0151] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
Claims
1. A path planning method for a drone swarm, characterized in that: The method comprises: Perform global path planning based on the path map within the movement range of the drone swarm to obtain the global path of the drone swarm; After the drone swarm starts to move based on the global path, the coordinates of the key points of the drone swarm at the current moment are determined in real time based on the positional relationship between the drones in the drone swarm; Determine multiple groups of local predicted trajectories according to the coordinates of key points of the drone swarm at the current moment, select local optimal trajectories from the multiple groups of local predicted trajectories according to the environmental obstacle information perceived by the drone swarm, locally optimize the global path based on the local optimal trajectory, and drive the drone swarm to continue moving based on the optimized global path; After the drone group starts to move based on the global path, the coordinates of the key points of the drone group at the current moment are determined in real time based on the positional relationship between the drones in the drone group, including: At the current moment, a first coordinate of a target UAV and a second coordinate of a neighboring UAV in a two-dimensional space are obtained, and a positioning diffusion degree between the target UAV and the neighboring UAV is calculated according to the first coordinate and the second coordinate; wherein the target UAV is any UAV in the UAV group, the neighboring UAV is any UAV in the UAV group except the target UAV, and the positioning diffusion degree is used to measure the distance between the target UAV and the neighboring UAV; Determine a diffusion matrix of the drone group based on the positioning diffusion between any two drones in the drone group, and calculate the proximity of the target drone based on the diffusion matrix; wherein the diffusion matrix is used to indicate the size of the distribution range of the drone group, and the proximity of the target drone indicates the degree of proximity between the first coordinate and all the second coordinates; Calculate the positioning mean and positioning error of the target UAV according to all approximations of the target UAV during its movement; The individual weight of the target UAV relative to the UAV swarm is calculated according to the positioning mean and positioning error of the target UAV, so as to obtain the individual weights of all UAVs relative to the UAV swarm; Calculate the coordinates of the key points of the drone group at the current moment according to the individual weights of all drones relative to the drone group, and the first coordinates and the second coordinates; The environmental obstacle information includes the distance between the local predicted trajectory and the obstacle. The method of determining multiple groups of local predicted trajectories according to the key point coordinates of the drone group at the current moment, and selecting the local optimal trajectory from the multiple groups of local predicted trajectories according to the environmental obstacle information perceived by the drone group, includes: The coordinates of the key points of the drone group at the current moment are used as the local starting point, a preset dynamic window is set with the local starting point as the starting reference point, and a local end point is determined within the preset dynamic window according to the global path; Simulating multiple groups of local predicted trajectories of the drone swarm between the local starting point and the local end point according to the current speed of the drone swarm; Scoring the multiple groups of local predicted trajectories according to the distances of the multiple groups of local predicted trajectories from the obstacles and a preset local path evaluation function, and selecting a group of local predicted trajectories with the highest score as the local optimal trajectory; wherein the local path evaluation function includes a static obstacle avoidance path scoring function and a dynamic obstacle avoidance path scoring function, the static obstacle avoidance path scoring function is the product of a first dynamic weight and a first distance, the dynamic obstacle avoidance path scoring function is the product of a second dynamic weight and a second distance, the first distance is the distance between a group of local predicted trajectories and the static obstacle, and the second distance is the distance between a group of local predicted trajectories and the dynamic obstacle; The global path planning is performed based on the path map within the moving range of the drone group to obtain the global path of the drone group, including: Connect the global starting point and the global end point set in the path map with a random route to generate the initial path of the drone swarm, and use the initial path of the drone swarm as the current path of the drone swarm; The current fitness value of the drone swarm is calculated based on a preset fitness function and the current path of the drone swarm, the current fitness value of the target drone is used as the current individual optimal value of the target drone, and the current minimum fitness value is used as the current group optimal value of the drone swarm; wherein the target drone is any drone in the drone swarm, and the current minimum fitness value is the minimum value among the current fitness values of the drone swarm; Update the current path of the drone swarm according to the current individual optimal value of the target drone and the current group optimal value of the drone swarm to obtain an updated path of the drone swarm; Determine whether the number of current update paths is equal to the preset maximum number of iterations; If the number of the current update path is less than the preset maximum number of iterations, the current fitness value of the drone swarm is calculated based on the preset fitness function and the update path of the drone swarm. If the updated current fitness value of the target drone is less than the current individual optimal value of the target drone, the current individual optimal value of the target drone is updated with the updated current fitness value of the target drone. If the updated current minimum fitness value is less than the current group optimal value of the drone swarm, the current group optimal value of the drone swarm is updated with the updated current minimum fitness value. Taking the updated path of the drone swarm as the current path of the drone swarm, returning to execute the step of updating the current path of the drone swarm according to the current individual optimal value of the target drone and the current group optimal value of the drone swarm to obtain the updated path of the drone swarm and subsequent steps; If the number of times the current path is updated is equal to the preset maximum number of iterations, the updated path of the drone swarm currently obtained is output as the global path.
2. The method according to claim 1, characterized in that: After simulating multiple groups of local predicted trajectories of the drone group between the local starting point and the local end point according to the current speed of the drone group, the method further includes: At the current moment, the first dynamic weight is set for the ratio of the number of static obstacles to all obstacles, and the second dynamic weight is set for the ratio of the number of dynamic obstacles to all obstacles.
3. The method according to claim 1, characterized in that The updating of the current path of the drone swarm according to the current individual optimal value of the target drone and the current group optimal value of the drone swarm includes: The drone swarm is divided into discoverers and followers according to the current individual optimal values of all drone swarms; wherein the number ratio between the discoverers and the followers is a preset ratio, and the current fitness values of the discoverers are all less than the current fitness values of the followers; Update the current path of the discoverer based on the security value; Update the current path of the follower according to the path of the drone corresponding to the current optimal value of the group and the path of the drone corresponding to the current worst value of the group; wherein the current worst value of the group is the maximum value among the current fitness values of the drone group; A preset number of drones are randomly selected from the drone group as early warning detectors, and the current path of the early warning detector is updated according to the path of the drone corresponding to the current optimal value of the group and the path of the drone corresponding to the current worst value of the group.
4. The method according to claim 1, characterized in that: The method further comprises: According to the global path, a fuzzy obstacle avoidance rule and a fuzzy guidance rule of the drone swarm are constructed; wherein the input of the fuzzy obstacle avoidance rule is the distance between the drone and the obstacle in each direction, and the output of the fuzzy obstacle avoidance rule is a preset fuzzy distance; the input of the fuzzy guidance rule is the angle and distance between the drone and the global end point, and the output of the fuzzy guidance rule is a preset fuzzy direction.
5. A path planning device for a drone swarm, characterized in that: The device comprises: The global planning module is used to perform global path planning based on the path map within the movement range of the drone swarm to obtain the global path of the drone swarm; A real-time key point acquisition module is used to determine the key point coordinates of the drone group at the current moment based on the positional relationship between each drone in the drone group in real time after the drone group starts to move based on the global path; A local optimization module is used to determine multiple groups of local predicted trajectories according to the key point coordinates of the drone group at the current moment, select local optimal trajectories from the multiple groups of local predicted trajectories according to the environmental obstacle information perceived by the drone group, locally optimize the global path based on the local optimal trajectory, and drive the drone group to continue moving based on the optimized global path; After the drone group starts to move based on the global path, the coordinates of the key points of the drone group at the current moment are determined in real time based on the positional relationship between the drones in the drone group, including: At the current moment, a first coordinate of a target UAV and a second coordinate of a neighboring UAV in a two-dimensional space are obtained, and a positioning diffusion degree between the target UAV and the neighboring UAV is calculated according to the first coordinate and the second coordinate; wherein the target UAV is any UAV in the UAV group, the neighboring UAV is any UAV in the UAV group except the target UAV, and the positioning diffusion degree is used to measure the distance between the target UAV and the neighboring UAV; Determine a diffusion matrix of the drone group based on the positioning diffusion between any two drones in the drone group, and calculate the proximity of the target drone based on the diffusion matrix; wherein the diffusion matrix is used to indicate the size of the distribution range of the drone group, and the proximity of the target drone indicates the degree of proximity between the first coordinate and all the second coordinates; Calculate the positioning mean and positioning error of the target UAV according to all approximations of the target UAV during its movement; The individual weight of the target UAV relative to the UAV swarm is calculated according to the positioning mean and positioning error of the target UAV, so as to obtain the individual weights of all UAVs relative to the UAV swarm; Calculate the coordinates of the key points of the drone group at the current moment according to the individual weights of all drones relative to the drone group, and the first coordinates and the second coordinates; The environmental obstacle information includes the distance between the local predicted trajectory and the obstacle. The method of determining multiple groups of local predicted trajectories according to the key point coordinates of the drone group at the current moment, and selecting the local optimal trajectory from the multiple groups of local predicted trajectories according to the environmental obstacle information perceived by the drone group, includes: The coordinates of the key points of the drone group at the current moment are used as the local starting point, a preset dynamic window is set with the local starting point as the starting reference point, and a local end point is determined within the preset dynamic window according to the global path; Simulating multiple groups of local predicted trajectories of the drone swarm between the local starting point and the local end point according to the current speed of the drone swarm; Scoring the multiple groups of local predicted trajectories according to the distances of the multiple groups of local predicted trajectories from the obstacles and a preset local path evaluation function, and selecting a group of local predicted trajectories with the highest score as the local optimal trajectory; wherein the local path evaluation function includes a static obstacle avoidance path scoring function and a dynamic obstacle avoidance path scoring function, the static obstacle avoidance path scoring function is the product of a first dynamic weight and a first distance, the dynamic obstacle avoidance path scoring function is the product of a second dynamic weight and a second distance, the first distance is the distance between a group of local predicted trajectories and the static obstacle, and the second distance is the distance between a group of local predicted trajectories and the dynamic obstacle; The global path planning is performed based on the path map within the moving range of the drone group to obtain the global path of the drone group, including: Connect the global starting point and the global end point set in the path map with a random route to generate the initial path of the drone swarm, and use the initial path of the drone swarm as the current path of the drone swarm; The current fitness value of the drone swarm is calculated based on a preset fitness function and the current path of the drone swarm, the current fitness value of the target drone is used as the current individual optimal value of the target drone, and the current minimum fitness value is used as the current group optimal value of the drone swarm; wherein the target drone is any drone in the drone swarm, and the current minimum fitness value is the minimum value among the current fitness values of the drone swarm; Update the current path of the drone swarm according to the current individual optimal value of the target drone and the current group optimal value of the drone swarm to obtain an updated path of the drone swarm; Determine whether the number of current update paths is equal to the preset maximum number of iterations; If the number of the current update path is less than the preset maximum number of iterations, the current fitness value of the drone swarm is calculated based on the preset fitness function and the update path of the drone swarm. If the updated current fitness value of the target drone is less than the current individual optimal value of the target drone, the current individual optimal value of the target drone is updated with the updated current fitness value of the target drone. If the updated current minimum fitness value is less than the current group optimal value of the drone swarm, the current group optimal value of the drone swarm is updated with the updated current minimum fitness value. Taking the updated path of the drone swarm as the current path of the drone swarm, returning to execute the step of updating the current path of the drone swarm according to the current individual optimal value of the target drone and the current group optimal value of the drone swarm to obtain the updated path of the drone swarm and subsequent steps; If the number of times the current path is updated is equal to the preset maximum number of iterations, the updated path of the drone swarm currently obtained is output as the global path.
6. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 4.
7. A path planning device for a drone swarm, comprising a memory and a processor, characterized in that: The memory stores a computer program, and when the computer program is executed by the processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 4.
Citation Information
Patent Citations
Multi-AUV self-adaptive target searching and obstacle avoidance method oriented to unknown environment
CN108594834A
Route control method and device of inspection robot, storage medium and inspection robot
CN110727278A
Multi-unmanned aerial vehicle cooperative coverage and obstacle avoidance flight path planning method
CN111811513A
