Unmanned aerial vehicle cluster path planning method based on A* and speed obstacle avoidance method
By combining A* and speed obstacle avoidance methods in the UAV cluster path planning, using lidar mapping and orca algorithms, local optimization and path conflict problems are solved, and safe and feasible path planning and interaction of UAV clusters are realized. flight.
Patent Information
- Application Number
- CN202410957784.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-17
- Publication Date
- 2024-11-08
AI Technical Summary
Existing drone cluster path planning is prone to local optimization and path conflict problems.
Adopt the UAV cluster path planning method based on A* and speed obstacle avoidance method. By building a cluster drone planning simulation platform, obtaining lidar data for map construction, using hybrid A* algorithm for global planning, and combining with orca algorithm for Local planning, update path points in real time to avoid collisions.
The safe and feasible global path planning of multiple drones in obstacle environments is realized, which avoids local optimality and path conflicts and ensures safe and interactive flight between drones.
Smart Images

Figure CN118913277A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of multi-robot path planning, and in particular relates to a method for unmanned aerial vehicle cluster path planning based on A* and speed obstacle avoidance method. Background Art
[0002] The rapid progress of science and technology has promoted the rapid development of basic technologies in the field of drones, such as microprocessor technology, sensor technology, communication technology and intelligent control technology. The increasing miniaturization and integration of the above technologies have greatly improved the functions and performance of small drones. Drones can be equipped with a variety of sensors, such as binocular cameras and lidar, GPS navigation, to obtain information about the surrounding environment and location information in the environment, and can also be equipped with processors to process data online. Drones have attracted the attention of scholars at home and abroad with their own flexibility, simplicity, and high maneuverability. Under this premise, the path planning technology of drone clusters has also been continuously developed, and due to its broad application prospects, researchers have been more comprehensively explored.
[0003] In recent years, drone swarms have received extensive attention in the civilian field, such as agricultural and forestry plant protection, road inspection, and coordinated lifting. In farmland and forest farms, drone swarms can complete agricultural monitoring, sowing, and drug application, in line with the current development direction of smart agriculture and precision agriculture promoted by the country. In disaster areas such as floods and forest fires, drone swarms can complete tasks such as disaster information collection and cluster formation forest fire fighting. At the same time, drone swarms are also widely used in the military field, and can perform combat tasks such as intelligence reconnaissance, battlefield surveillance, target tracking, and military strikes. In a battlefield environment, drone swarms can collect intelligence and carry out precise strikes, such as "decapitation" operations. Unmanned swarm systems are a type of high-efficiency unmanned combat equipment that enhances national defense force construction. Major research centers around the world are planning key technology research on unmanned swarm systems from aspects such as system architecture and autonomous collaboration to achieve intelligent, collaborative, and efficient swarm combat.
[0004] Unmanned swarm technology involves technologies such as situation perception, path planning, communication network and formation control. Cluster path planning is an important part of unmanned swarm technology. Cluster path planning technology is developed from single drone planning technology. It still has some characteristics of single drone and has further expanded these technologies. After the local perception information and information interaction obtained by the unmanned system, the individual drone realizes obstacle avoidance motion path planning with both autonomy and real-time, and then the controller is used to track the planned path. The unmanned swarm adopts a distributed architecture. Compared with the centralized unmanned swarm framework, it does not require centralized control of the central control unit. Various parameters can be transmitted between robots through communication, so as to achieve mutual coordination to jointly complete the predetermined tasks. The system can make the system have better scalability and fault tolerance. However, the existing unmanned cluster path planning is prone to local optimality and path conflict problems. Based on this, the present invention proposes a drone cluster path planning method based on A* and speed obstacle avoidance method. Summary of the invention
[0005] In view of the above technical problems, the present invention provides a UAV cluster path planning method based on A* and speed obstacle avoidance method.
[0006] The technical solution adopted by the present invention to solve the technical problem is:
[0007] A method for UAV cluster path planning based on A* and speed obstacle avoidance method, the method comprising the following steps:
[0008] S100: Build a cluster drone planning simulation platform, including the simulation platform and simulation model files, drone SDF model files and laser radar camera model files located above the drone, a plug-in for obtaining global drone position information, and a px4 flight controller for completing drone flight control;
[0009] S200: acquiring laser radar data, and obtaining an occupancy grid map for planning through a laser radar mapping algorithm;
[0010] S300: After obtaining the position and speed information of the UAV in the simulation and the occupancy grid map information updated in real time in S200, the hybrid A* algorithm is used to perform global planning of the UAV;
[0011] S400: Each drone receives path information from the global planning nodes of all other drones, and uses the orca local planning algorithm to complete the path planning within the local range of the current drone, taking the next global path point as the target point to achieve drone interaction and flight, and complete drone cluster flight and collision avoidance.
[0012] Preferably, S100 includes:
[0013] S110: Use the Gazebo simulation platform under ROS to build a virtual simulation, start the drone SDF model file, and start px4 to establish information communication between the simulation platform and the flight control;
[0014] S120: Start the laser radar camera model file and the drone position plug-in at the same time to obtain the true value of the position in the simulation platform, which is used as the global position information of the drone.
[0015] Preferably, S200 includes:
[0016] S210: Obtain the point cloud data of the laser radar, start the octomap node, and use the octomap algorithm to build a map to convert the point cloud map to the occupancy grid map;
[0017] S220: Send the converted occupancy grid map to the drone global planning node, and update the occupancy grid map in real time, and display it in the visualization software rviz.
[0018] Preferably, S300 includes:
[0019] S310: Start the hybrid A* algorithm node, receive the occupancy grid map information sent by the octomap node, delete the point cloud information of other identified drones, and then expand the map;
[0020] S320: Subscribe to the true value information of the UAV's position and posture from the simulation platform, set the starting point and the target point, and when the UAV reaches a certain height and stabilizes, use the hybrid A* algorithm to search the UAV's global path to obtain a safe and feasible UAV global path;
[0021] S330: Send the searched safe and feasible global path of the UAV to the local planning node of the UAV, and update the global path information while the position of the UAV is constantly moving.
[0022] Preferably, S320 specifically includes:
[0023] S321: Initialize the starting point and the target point, initialize the open list and the close list, and add the starting point to the open list;
[0024] S322: Select the state node with the smallest cost in the open list as the current node, and delete the current node from the open list;
[0025] S323: Determine whether the current node is the target node. If the current node is the target node, end the return path; if the current node is not the target node, execute S324;
[0026] S324: add the current node to the close list, search and expand the current node, generate all child nodes of the current node, and judge each child node. If the child node already exists in the close list, skip the child node; if the child node is not in the open list, set the current node as the parent node of the child node, calculate the cost of the child node, and add it to the open list; if the child node is already in the open list, update the cost of the child node, and take the better cost to add it to the open list;
[0027] S325: Repeat S322 to S324. If the open list is empty, the search fails and a null value is returned.
[0028] Preferably, every preset number of cycles, S323 further includes:
[0029] If the current node is not the target node, the Reeds-Sheep curve is used to generate a trajectory from the current node to the target point, and it is determined whether the trajectory falls into the expanded occupancy grid map. If not, it means that no collision has occurred, and the result returns the path. If a collision occurs, S324 is executed.
[0030] Preferably, the cost refers to the sum of the path length from the starting state to the current state and the distance from the current state to the target state estimated by the heuristic function. S322 is specifically:
[0031] f(n)=g(n)+h(n)
[0032] Where f(n) is the sum of the costs, g(n) is the path length from the starting state to the current state, and h(n) is the distance from the current state to the target state estimated by the heuristic function.
[0033] Preferably, S400 includes:
[0034] S410: Start the orca algorithm node, receive the safe and feasible global path information of the UAV sent from the hybrid A* node, divide a path into several segments evenly, and add the selected path points to the target point sequence of the orca algorithm;
[0035] S420: Subscribe to the true value information of the drone's position and attitude from the simulation platform. Each drone orca node receives the position and speed information from other drone orca nodes, and sends out its own position and speed information.
[0036] S430: During the movement toward the target point, the position and speed of the UAV are updated in real time, the global path points are updated, and the local planning of the UAV is completed.
[0037] Preferably, the multi-UAV orca algorithm in S410 is specifically:
[0038] S411: Perceiving the position, shape and speed of other drones: The position P, radius R and speed V of a drone are described as the external properties of the drone, which can be observed by other drones;
[0039] S412: Calculate the orca between drones;
[0040] S413: After the drones calculate each other's orca, an optional new speed intersection is obtained, and a new speed closest to the preset expected speed is selected from the optional new speed intersection; wherein the expected speed and the new speed of the drone are described as internal attributes of the drone;
[0041] S414: Use the obtained new speed to update the position of the UAV, and then return to execute S411.
[0042] Preferably, S412 is specifically:
[0043] Assume there are two drones A and B. Assume drone B is stationary at this time. Set a certain time τ and calculate the speed obstacle area of drones A and B. It is expressed as follows:
[0044]
[0045] The region has a geometric shape of a truncated cone, so the set of velocities where no collision occurs is The complement of
[0046] Assume that B's selection speed set is V B , then we can find the speed set of drones A and B that will not collide within τ time It is expressed as follows:
[0047]
[0048] Expressed as With V B The complement of the Minkowski sum of , Denotes the Minkowski sum, assuming that A and B use speed and Such a speed causes A and B to collide, that is,
[0049] Let u be The vector to the closest point on the velocity obstacle boundary, u, is expressed as follows:
[0050]
[0051] u is the value that minimizes the change in the speed of A and B within the time τ in order to avoid collision. is the symbol of partial derivative;
[0052] Let n be On the border In order to fairly "share" the responsibility of avoiding collisions between robots, robot A should at least Adjust your own speed and assume that B is responsible for the other half, so the set The allowed speed is based on the point The half-plane starting from n is a half-plane cut by a straight line, which is perpendicular to u. The formula is as follows:
[0053]
[0054] In view of the problem that local planning in the existing technology will fall into the local optimum, and the path conflict problem will occur in the global path planning of multiple drones, a drone cluster path planning method based on A* and speed obstacle avoidance method is proposed, which can build maps in real time, complete the global path planning and local planning of multiple drones, and thus complete the interactive flight and counter-flight of multiple drones. The global path planning algorithm is used to provide global path points for the local planning algorithm to prevent drones from falling into the local optimum for a long time. The local path planning algorithm is added to solve the interaction problem between drones and avoid collisions between drones. BRIEF DESCRIPTION OF THE DRAWINGS
[0055] Figure 1 This is a flow chart of a method for UAV cluster path planning based on A* and speed obstacle avoidance method of the present invention;
[0056] Figure 2 It is a framework diagram of a UAV cluster path planning method based on A* and speed obstacle avoidance method of the present invention;
[0057] Figure 3 is a flow chart of the hybrid A* algorithm of the present invention;
[0058] Figure 4 It is the flow chart of the orca algorithm of the present invention;
[0059] Figure 5 It is the simulation working environment of the present invention;
[0060] Figure 6 It is a completed example of the UAV path planning of the present invention. DETAILED DESCRIPTION
[0061] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings.
[0062] In one embodiment, Figure 1 and 2 As shown, a UAV cluster path planning method based on A* and speed obstacle avoidance method, the method comprises the following steps:
[0063] S100: Build a cluster drone planning simulation platform, including the simulation platform and simulation model files, drone SDF model files and laser radar camera model files located above the drone, a plug-in for obtaining global drone position information, and a px4 flight controller for completing drone flight control;
[0064] S200: acquiring laser radar data, and obtaining an occupancy grid map for planning through a laser radar mapping algorithm;
[0065] S300: After obtaining the position and speed information of the UAV in the simulation and the occupancy grid map information updated in real time in S200, the hybrid A* algorithm is used to perform global planning of the UAV;
[0066] S400: Each drone receives path information from the global planning nodes of all other drones, and uses the orca local planning algorithm to complete the path planning within the local range of the current drone, taking the next global path point as the target point to achieve drone interaction and flight, and complete drone cluster flight and collision avoidance.
[0067] Aiming at the problem that local planning in the prior art may fall into local optimality and that global path planning of multiple UAVs may have path conflict problems, a UAV cluster path planning method based on A* and speed obstacle avoidance method is proposed, which includes the steps of building a cluster UAV planning simulation platform, a laser radar mapping step, a global path planning step and a local path planning step. The cluster UAV planning simulation platform building step is an SDF model file containing multiple UAV models and a laser radar model for local mapping. The laser radar mapping step is to convert the point cloud map recognized by the laser radar sensor into an occupancy grid map. The global path planning step is to generate a collision-free path in real time within the local map range after subscribing to the occupancy grid map. The local path planning step is to avoid other UAVs and obstacles within the local range after subscribing to the globally planned path. The invention can enable multiple UAVs to search for safe and feasible paths in an obstacle environment and avoid collisions with each other.
[0068] In one embodiment, S100 includes:
[0069] S110: Use the Gazebo simulation platform under ROS to build a virtual simulation, start the drone SDF model file, and start px4 to establish information communication between the simulation platform and the flight control;
[0070] S120: Start the laser radar camera model file and the drone position plug-in at the same time to obtain the true value of the position in the simulation platform, which is used as the global position information of the drone.
[0071] In one embodiment, S200 includes:
[0072] S210: Obtain the point cloud data of the laser radar, start the octomap node, and use the octomap algorithm to build a map to convert the point cloud map to the occupancy grid map;
[0073] S220: Send the converted occupancy grid map to the drone global planning node, and update the occupancy grid map in real time, and display it in the visualization software rviz.
[0074] In one embodiment, S300 includes:
[0075] S310: Start the hybrid A* algorithm node, receive the occupancy grid map information sent by the octomap node, delete the point cloud information of other identified drones, and then expand the map;
[0076] S320: Subscribe to the true value information of the UAV's position and posture from the simulation platform, set the starting point and the target point, and when the UAV reaches a certain height and stabilizes, use the hybrid A* algorithm to search the UAV's global path to obtain a safe and feasible UAV global path;
[0077] S330: Send the searched safe and feasible global path of the UAV to the local planning node of the UAV, and update the global path information while the position of the UAV is constantly moving.
[0078] Specifically, the expansion of the map in S310 refers to setting the expansion radius, and expanding the acquired obstacle point cloud map outward according to the expansion radius to prevent the UAV from colliding with obstacles; the true value information of the UAV's posture in S320 includes the position, speed and acceleration information of the UAV; the safe and feasible global path of the UAV in S330 means that the globally planned path is not in the obstacle, that is, it does not fall into the expanded occupancy grid map.
[0079] In one embodiment, Figure 3 As shown, S320 is specifically:
[0080] S321: Initialize the starting point and the target point, initialize the open list and the close list, and add the starting point to the open list;
[0081] S322: Select the state node with the smallest cost in the open list as the current node, and delete the current node from the open list;
[0082] S323: Determine whether the current node is the target node. If the current node is the target node, end the return path; if the current node is not the target node, execute S324;
[0083] S324: add the current node to the close list, search and expand the current node, generate all child nodes of the current node, and judge each child node. If the child node already exists in the close list, skip the child node; if the child node is not in the open list, set the current node as the parent node of the child node, calculate the cost of the child node, and add it to the open list; if the child node is already in the open list, update the cost of the child node, and take the better cost to add it to the open list;
[0084] S325: Repeat S322 to S324. If the open list is empty, the search fails and a null value is returned.
[0085] Specifically, in S323, if the current node is the target node, the search is completed. The current node has a parent node, and the path can be backtracked from the current node to the initial node to search for the parent node. Therefore, the returned path is a path from the starting node to the target node.
[0086] In one embodiment, every preset number of cycles, S323 further includes:
[0087] If the current node is not the target node, the Reeds-Sheep curve is used to generate a trajectory from the current node to the target node, and it is determined whether the trajectory falls into the expanded occupancy grid map. If not, it means that no collision has occurred, and the result returns the path. If a collision occurs, S324 is executed.
[0088] Specifically, if the current node is not the target node, the current node is directly connected to the target node through a curve to determine whether the trajectory is within the obstacle range, saving the time of continuing to search for adjacent nodes. This strategy is set to be executed once every preset number of cycles.
[0089] In one embodiment, the cost refers to the sum of the path length from the starting state to the current state and the distance from the current state to the target state estimated by the heuristic function. S322 is specifically:
[0090] f(n)=g(n)+h(n)
[0091] Where f(n) is the sum of the costs, g(n) is the path length from the starting state to the current state, and h(n) is the distance from the current state to the target state estimated by the heuristic function.
[0092] In one embodiment, S400 includes:
[0093] S410: Start the orca algorithm node, receive the safe and feasible global path information of the UAV sent from the hybrid A* node, divide a path into several segments evenly, and add the selected path points to the target point sequence of the orca algorithm;
[0094] S420: Subscribe to the true value information of the drone's position and attitude from the simulation platform. Each drone orca node receives the position and speed information from other drone orca nodes, and sends out its own position and speed information.
[0095] S430: During the movement toward the target point, the position and speed of the UAV are updated in real time, the global path points are updated, and the local planning of the UAV is completed.
[0096] Specifically, in S410, path points are selected at average intervals and added to the target point sequence of the orca algorithm, but the selection is not limited to this method.
[0097] In one embodiment, Figure 4 As shown, the multi-UAV orca algorithm in S410 is specifically as follows:
[0098] S411: Perceiving the position, shape and speed of other drones: The position P, radius R and speed V of a drone are described as the external properties of the drone, which can be observed by other drones;
[0099] S412: Calculate the orca between drones;
[0100] S413: After the drones calculate each other's orca, an optional new speed intersection is obtained, and a new speed closest to the preset expected speed is selected from the optional new speed intersection; wherein the expected speed and the new speed of the drone are described as internal attributes of the drone;
[0101] S414: Use the obtained new speed to update the position of the UAV, and then return to execute S411.
[0102] In one embodiment, S412 specifically includes:
[0103] Assume there are two drones A and B. Assume drone B is stationary at this time. Set a certain time τ and calculate the speed obstacle area of drones A and B. It is expressed as follows:
[0104]
[0105] The region has a geometric shape of a truncated cone, so the set of velocities where no collision occurs is The complement of
[0106] Assume that B's selection speed set is V B , then we can find the speed set of drones A and B that will not collide within τ time It is expressed as follows:
[0107]
[0108] Expressed as With V B The complement of the Minkowski sum of , Denotes the Minkowski sum, assuming that A and B use speed and Such a speed causes A and B to collide, that is,
[0109] Let u be The vector to the closest point on the velocity obstacle boundary, u, is expressed as follows:
[0110]
[0111] u is the value that minimizes the change in the speed of A and B within the time τ in order to avoid collision. is the symbol of partial derivative;
[0112] Let n be On the border In order to fairly "share" the responsibility of avoiding collisions between robots, robot A should at least Adjust your own speed and assume that B is responsible for the other half, so the set The allowed speed is based on the point The half-plane starting from n is a half-plane cut by a straight line, which is perpendicular to u. The formula is as follows:
[0113]
[0114] Specifically, similarly, robot B The same is true for the allowed speeds. Each drone will have an optional speed set, from which a new speed closest to the preset desired speed is selected; the new speed is used to update the drone's position, and then the above steps are continued.
[0115] like Figure 5 and 6 As shown, this example is run in the experimental platform to verify the feasibility and optimality of the present invention. It can be understood that the present invention is not limited to this example, and the present invention can be expanded by increasing the number of drones, adding new environments, etc.
[0116] The above-mentioned UAV cluster path planning method based on A* and speed obstacle avoidance method has the following advantages:
[0117] 1. In order to solve the problem that local planning may fall into local optimality, a global path planning algorithm is used to provide global path points for the local planning algorithm to prevent the drone from falling into local optimality for a long time. When there are no other drones in the local area, the path points of global path planning are used to move the drone.
[0118] 2. In view of the path conflict problem that may occur in the global path planning of multiple drones, a local path planning algorithm is added to solve the interaction problem between drones to avoid collisions between drones. When there are other drones in the local area, orca local planning is used to complete the interaction of drones.
[0119] 3. Real-time environmental perception and mapping are realized. The global point cloud information is obtained by using lidar, which is converted into an occupancy grid map through the octomap algorithm. After deleting the map information of other drones, the occupancy grid map is expanded to meet the requirements of the drone's global path planning.
[0120] The above is a detailed introduction to the UAV cluster path planning method based on A* and speed obstacle avoidance method provided by the present invention. This article uses specific examples to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the core idea of the present invention. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present invention, the present invention can also be improved and modified, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.
Claims
1. A UAV cluster path planning method based on A* and speed obstacle avoidance method, characterized in that: The method comprises the following steps: S100: Build a cluster drone planning simulation platform, including the simulation platform and simulation model files, drone SDF model files and laser radar camera model files located above the drone, a plug-in for obtaining global drone position information, and a px4 flight controller for completing drone flight control; S200: acquiring laser radar data, and obtaining an occupancy grid map for planning through a laser radar mapping algorithm; S300: After obtaining the position and speed information of the UAV in the simulation and the occupancy grid map information updated in real time in S200, the hybrid A* algorithm is used to perform global planning of the UAV; S400: Each drone receives path information from the global planning nodes of all other drones, and uses the orca local planning algorithm to complete the path planning within the local range of the current drone, taking the next global path point as the target point to achieve drone interaction and flight, and complete drone cluster flight and collision avoidance.
2. The method according to claim 1, characterized in that S100 includes: S110: Use the Gazebo simulation platform under ROS to build a virtual simulation, start the drone SDF model file, and start px4 to establish information communication between the simulation platform and the flight control; S120: Start the laser radar camera model file and the drone position plug-in at the same time to obtain the true value of the position in the simulation platform, which is used as the global position information of the drone.
3. The method according to claim 2, characterized in that S200 includes: S210: Obtain the point cloud data of the laser radar, start the octomap node, and use the octomap algorithm to build a map to convert the point cloud map to the occupancy grid map; S220: Send the converted occupancy grid map to the drone global planning node, and update the occupancy grid map in real time, and display it in the visualization software rviz.
4. The method according to claim 3, characterized in that S300 includes: S310: Start the hybrid A* algorithm node, receive the occupancy grid map information sent by the octomap node, delete the point cloud information of other identified drones, and then expand the map; S320: Subscribe to the true value information of the UAV's position and posture from the simulation platform, set the starting point and the target point, and when the UAV reaches a certain height and stabilizes, use the hybrid A* algorithm to search the UAV's global path to obtain a safe and feasible UAV global path; S330: Send the searched safe and feasible global path of the UAV to the local planning node of the UAV, and update the global path information while the position of the UAV is constantly moving.
5. The method according to claim 4, characterized in that S320 is specifically: S321: Initialize the starting point and the target point, initialize the open list and the close list, and add the starting point to the open list; S322: Select the state node with the smallest cost in the open list as the current node, and delete the current node from the open list; S323: Determine whether the current node is the target node. If the current node is the target node, end the return path. If the current node is not the target node, execute S324; S324: add the current node to the close list, search and expand the current node, generate all child nodes of the current node, and judge each child node. If the child node already exists in the close list, skip the child node; If the child node is not in the open list, set the current node as the parent node of the child node, calculate the cost of the child node, and add it to the open list; If the child node is already in the open list, update the cost of the child node and add the node with a better cost to the open list; S325: Repeat S322 to S324. If the open list is empty, the search fails and a null value is returned.
6. The method according to claim 5, characterized in that Every preset number of cycles, S323 also includes: If the current node is not the target node, the Reeds-Sheep curve is used to generate a trajectory from the current node to the target point, and it is determined whether the trajectory falls into the expanded occupancy grid map. If not, it means that no collision has occurred, and the result returns the path. If a collision occurs, S324 is executed.
7. The method according to claim 6, characterized in that The cost refers to the sum of the path length from the starting state to the current state and the distance from the current state to the target state estimated by the heuristic function. S322 is specifically: f(n)=g(n)+h(n) Where f(n) is the sum of the costs, g(n) is the path length from the starting state to the current state, and h(n) is the distance from the current state to the target state estimated by the heuristic function.
8. The method according to claim 7, characterized in that S400 includes: S410: Start the orca algorithm node, receive the safe and feasible global path information of the UAV sent from the hybrid A* node, divide a path into several segments evenly, and add the selected path points to the target point sequence of the orca algorithm; S420: Subscribe to the true value information of the drone's position and attitude from the simulation platform. Each drone orca node receives the position and speed information from other drone orca nodes, and sends out its own position and speed information. S430: During the movement toward the target point, the position and speed of the UAV are updated in real time, the global path points are updated, and the local planning of the UAV is completed.
9. The method according to claim 8, characterized in that The specific orca algorithm of multiple drones in S410 is: S411: Perceiving the position, shape and speed of other drones: The position P, radius R and speed V of a drone are described as the external properties of the drone, which can be observed by other drones; S412: Calculate the orca between drones; S413: After the drones calculate each other's orca, an optional new speed intersection is obtained, and a new speed closest to the preset expected speed is selected from the optional new speed intersection; wherein the expected speed and the new speed of the drone are described as internal attributes of the drone; S414: Use the obtained new speed to update the position of the UAV, and then return to execute S411.
10. The method according to claim 9, characterized in that S412 is specifically: Assume there are two drones A and B. Assume drone B is stationary at this time. Set a certain time τ and calculate the speed obstacle area of drones A and B. It is expressed as follows: The region has a geometric shape of a truncated cone, so the set of velocities where no collision occurs is The complement of Assume that B's selection speed set is V B , then we can find the speed set of drones A and B that will not collide within τ time It is expressed as follows: Expressed as With V B The complement of the Minkowski sum of , Denotes the Minkowski sum, assuming that A and B use speed and Such a speed causes A and B to collide, that is, Let u be The vector to the closest point on the velocity obstacle boundary, u, is expressed as follows: u is the value that minimizes the change in the speed of A and B within the time τ in order to avoid collision. is the symbol of partial derivative; Let n be On the border In order to fairly "share" the responsibility of avoiding collisions between robots, robot A should at least Adjust your own speed and assume that B is responsible for the other half, so the set The allowed speed is based on the point The half-plane starting from n is a half-plane cut by a straight line, which is perpendicular to u. The formula is as follows:
Citation Information
Cited By
Unmanned aerial vehicle cluster distribution planning method based on deep learning
CN121113094A
Intensive goods allocation warehouse-in and warehouse-out method and system based on MAP map and medium
CN121212941A
