A flight control method, device, medium and product for a multi-UAV cluster

Abstract

A flight control method, device, medium, and product for a multi-UAV swarm, relating to the general field of control or regulation systems, includes: obtaining the main UAV's position data and a preset flight mission, and broadcasting the main UAV's position data to the surrounding space; receiving neighboring position data in multiple search directions, determining the number of neighboring UAVs, and calculating a main UAV density value; obtaining neighboring density values ​​of neighboring UAVs in multiple search directions and determining obstacle locations; determining an obstacle avoidance direction and a reverse direction based on the main UAV density and the neighboring density values; obtaining an obstacle avoidance rehearsal space for neighboring UAVs in the reverse direction, determining an obstacle avoidance adjustment path, and calculating a flight path for the main UAV; and generating flight control instructions for the main UAV based on the flight path to control the main UAV to execute the preset flight mission. Implementing this application can optimize the task execution efficiency of a multi-UAV swarm in high-density formations.

Description

Technical Field

[0001] The present application relates to the general field of control or regulation systems, and in particular to a flight control method, device, medium and product for a multi-UAV cluster. Background Art

[0002] With the rapid development of drone technology, drone swarms are increasingly being used in military reconnaissance, emergency rescue, environmental monitoring, and other fields. When performing missions in complex and dynamic environments, drone swarms require flexible control mechanisms to respond to various emergencies and ensure reliable mission completion. At the same time, the increasing complexity of mission environments and the scale of swarms place higher demands on the coordination, robustness, and adaptability of control systems.

[0003] In related technologies, drone swarm control primarily utilizes centralized or distributed control schemes. Centralized control utilizes a ground control station or a master drone to centrally manage the entire swarm, responsible for mission planning, path planning, and command issuance. Distributed control, on the other hand, allows each drone to make autonomous decisions based on local perception and pre-set rules, achieving swarm coordination through local communication. These two control schemes, each tailored to specific application scenarios, have demonstrated promising results in practice.

[0004] However, when multiple drones form a high-density formation, if multiple drones need to avoid obstacles at the same time, the obstacle avoidance actions of each drone will affect the flight space of other drones, forcing more drones to change their flight paths. Especially in a small space, the change of the collective flight path will significantly reduce the efficiency of mission execution. Summary of the Invention

[0005] The present application provides a flight control method, device, medium and product for a multi-UAV cluster, which are used to optimize the mission execution efficiency of a multi-UAV cluster in high-density formation situations.

[0006] In a first aspect, the present application provides a flight control method for a multi-UAV cluster, which is applied to a control device, which is arranged on each UAV in the multi-UAV cluster, and the method includes: obtaining the main body position data and preset flight mission of the main body UAV, and broadcasting the main body position data to the surrounding space; receiving adjacent position data in multiple search directions, determining the number of adjacent UAVs, and calculating the main body density value of the main body UAV in the search direction; the adjacent UAV is a UAV within a preset distance range of the main body UAV in the search direction; obtaining the adjacent density values ​​of the adjacent UAVs in multiple search directions, and determining the obstacle position when the main body UAV detects an obstacle; determining an obstacle avoidance direction and an obstacle avoidance reverse direction based on the main body density value and the adjacent density value; the obstacle avoidance direction and the obstacle avoidance reverse direction are opposite, and the obstacle avoidance direction and the obstacle avoidance reverse direction are both one of the multiple search directions; obtaining an obstacle avoidance rehearsal space of the adjacent UAV in the obstacle avoidance reverse direction; determining an obstacle avoidance adjustment path based on the obstacle avoidance rehearsal space, the obstacle position, the obstacle avoidance direction and the preset flight mission, and calculating the flight path of the main body UAV; generating a flight control instruction for the main body UAV based on the flight path to control the main body UAV to perform the preset flight mission.

[0007] In the above embodiment, the control device realizes distributed obstacle avoidance control based on local density by obtaining its own position data and broadcasting and receiving the position data of nearby drones, and calculating the density value to determine the obstacle avoidance direction; it enables each drone to independently perceive the surrounding environment and autonomously plan the obstacle avoidance path, while ensuring safety while maintaining the execution efficiency of the original flight mission to the greatest extent, effectively solving the obstacle avoidance coordination problem in high-density formation conditions.

[0008] In combination with some embodiments of the first aspect, in some embodiments, the step of determining the obstacle avoidance direction and the opposite direction of obstacle avoidance based on the main body density value and the adjacent density value specifically includes: calculating the density difference in multiple search directions based on the main body density value and the adjacent density value; the density difference is the difference between the main body density value and the adjacent density value in the corresponding search direction; selecting the search direction with the largest density difference as the obstacle avoidance direction, and determining the search direction opposite to the obstacle avoidance direction as the opposite direction of obstacle avoidance.

[0009] In the above embodiment, the control device calculates the density difference based on the main body density value and the adjacent density value, and selects the direction with the largest density difference as the obstacle avoidance direction, so that the UAV can avoid obstacles in the direction of lower density, which can minimize the impact on other UAVs, reduce the probability of chain obstacle avoidance, and improve the maneuverability of the entire cluster.

[0010] In combination with some embodiments of the first aspect, in some embodiments, the step of obtaining the obstacle avoidance rehearsal space of the adjacent UAV in the opposite direction of obstacle avoidance specifically includes: sending an obstacle avoidance rehearsal request to the adjacent UAV in the opposite direction of obstacle avoidance; receiving obstacle avoidance coordination data returned by the adjacent UAV in the opposite direction of obstacle avoidance; and calculating the obstacle avoidance rehearsal space based on the obstacle position and the obstacle avoidance coordination data.

[0011] In the above embodiment, the control device realizes obstacle avoidance space coordination between drones by sending a rehearsal request to the adjacent drone in the opposite direction of obstacle avoidance and receiving coordination data, so that the main drone can accurately grasp the availability of obstacle avoidance space, plan a suitable obstacle avoidance path in advance, and avoid the risk of collision caused by blind obstacle avoidance.

[0012] In combination with some embodiments of the first aspect, in some embodiments, the step of calculating the obstacle avoidance rehearsal space based on the obstacle position and obstacle avoidance coordination data specifically includes: determining the current flight status and expected flight trajectory of the adjacent UAV in the opposite direction of the obstacle avoidance based on the obstacle avoidance coordination data; calculating the available obstacle avoidance space of the main UAV in the direction of travel based on the current flight status, expected flight trajectory and obstacle position; determining the obstacle avoidance rehearsal space based on the available obstacle avoidance space and the maneuverability constraints of the main UAV.

[0013] In the above embodiment, the control device calculates the available obstacle avoidance space based on the current state and expected trajectory of the adjacent UAV, and determines the obstacle avoidance rehearsal space in combination with the maneuverability constraints of the main UAV, thereby ensuring the feasibility of the obstacle avoidance path and improving the accuracy and reliability of the obstacle avoidance action.

[0014] In combination with some embodiments of the first aspect, in some embodiments, before the steps of receiving adjacent position data in multiple search directions, determining the number of machines adjacent to the drone, and calculating the body density value of the main drone in the search direction, the method also includes: obtaining environmental parameters in multiple search directions; the environmental parameters include wind speed, wind direction, air pressure and temperature; assigning weights to multiple search directions based on the environmental parameters to obtain search direction weight coefficients; adjusting the effective detection distances in multiple search directions according to the search direction weight coefficients, and determining a preset distance range based on the effective detection distances.

[0015] In the above embodiment, the control device achieves dynamic adjustment of the detection distance in different search directions by acquiring environmental parameters and assigning weights, so that the UAV can more accurately assess the surrounding environmental conditions, improve the accuracy of density calculation and the rationality of obstacle avoidance decisions.

[0016] In combination with some embodiments of the first aspect, in some embodiments, after determining the obstacle avoidance adjustment path based on the obstacle avoidance rehearsal space, obstacle position, obstacle avoidance direction and preset flight mission, and calculating the flight path of the main body UAV, the method also includes: detecting dynamic obstacles on the flight path and calculating the preset trajectory of the dynamic obstacles; based on the preset trajectory, updating the obstacle avoidance adjustment path and the flight path.

[0017] In the above embodiment, the control device realizes the recognition and prediction of dynamic obstacles and updates the obstacle avoidance path based on the predicted trajectory, so that the UAV can cope with complex and changing environments, and improves the environmental adaptability and mission reliability of the system.

[0018] In combination with some embodiments of the first aspect, in some embodiments, the step of updating the obstacle avoidance adjustment path and the flight path based on the preset trajectory specifically includes: constructing a dynamic obstacle avoidance area based on the preset trajectory of the dynamic obstacle; calculating the minimum safe distance between the main UAV and the dynamic obstacle avoidance area, and determining the dynamic obstacle avoidance priority based on the minimum safe distance; adjusting the obstacle avoidance strategy according to the dynamic obstacle avoidance priority and the real-time status of the obstacle avoidance rehearsal space; updating the obstacle avoidance adjustment path based on the obstacle avoidance strategy, and replanning the flight path in combination with the preset flight mission.

[0019] In the above embodiment, the control device realizes the priority-based dynamic obstacle avoidance strategy adjustment by constructing a dynamic obstacle avoidance area and calculating the minimum safety distance, thereby ensuring the obstacle avoidance effect when multiple dynamic obstacles exist at the same time and improving the safety and robustness of the system.

[0020] In a second aspect, an embodiment of the present application provides a control device, comprising: one or more processors and a memory; the memory is coupled to the one or more processors, the memory being used to store computer program code, the computer program code comprising computer instructions, the one or more processors calling the computer instructions to enable the control device to execute the method described in the first aspect and any possible implementation of the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer program product comprising instructions, which, when the computer program product is run on a control device, enables the control device to execute the method described in the first aspect and any possible implementation of the first aspect.

[0022] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium comprising instructions, which, when executed on a control device, causes the control device to execute the method described in the first aspect and any possible implementation of the first aspect.

[0023] It is understandable that the control device provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the methods provided in the embodiments of the present application. Therefore, the beneficial effects that can be achieved can be referred to the beneficial effects of the corresponding methods and will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0025] 1. Due to the adoption of a distributed obstacle avoidance control method based on the density value of the main body and the density value of the neighboring drones, each drone can make independent decisions based on local perception information and achieve collaborative obstacle avoidance through density calculation and obstacle avoidance rehearsal. Therefore, drones can select the optimal obstacle avoidance path while ensuring flight safety. This effectively solves the problems of system response delay and excessive communication burden caused by centralized control in related technologies, thereby achieving efficient obstacle avoidance and mission execution in high-density formations.

[0026] 2. Due to the adoption of a spatial rehearsal mechanism based on obstacle avoidance rehearsal requests and collaborative data, by sending rehearsal requests to nearby drones and receiving their flight status and expected trajectory information, the main drone can accurately assess the available obstacle avoidance space. Therefore, the planning of obstacle avoidance paths is more forward-looking and feasible, effectively solving the problems of blind obstacle avoidance and chain obstacle avoidance caused by incomplete information in related technologies, thereby achieving a smoother and more efficient obstacle avoidance process; it not only ensures the safety of obstacle avoidance actions, but also improves the maneuverability of the entire cluster by minimizing the impact on other drones.

[0027] 3. Due to the use of dynamic obstacle recognition and trajectory prediction technology, combined with real-time detection and preset trajectory calculation, the UAV system can predict the movement trend of obstacles in advance and adjust the obstacle avoidance strategy in time. Therefore, it has stronger adaptability when facing complex and changing environments, effectively solving the safety hazard caused by insufficient response to dynamic obstacles in related technologies, and thus achieving more flexible and reliable obstacle avoidance control. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a flow chart of a flight control method for a multi-UAV cluster in an embodiment of the present application;

[0029] Figure 2 This is another flowchart of the flight control method for a multi-UAV cluster according to an embodiment of the present application;

[0030] Figure 3 It is a schematic diagram of the structure of a physical device of the control device in the embodiment of the present application. DETAILED DESCRIPTION

[0031] The terms used in the following examples of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification of the present application, the singular expressions "a", "an", "above", "the", and "this" are intended to include plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations of one or more of the listed items.

[0032] In the following, the terms "first" and "second" are used for descriptive purposes only and should not be understood to imply or suggest relative importance or implicitly indicate the number of the technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of this application, unless otherwise specified, "plurality" means two or more.

[0033] For ease of understanding, the application scenarios of the embodiments of the present application are introduced below.

[0034] A large urban logistics distribution center handles over 10,000 express delivery orders daily, covering multiple residential and commercial areas within a 15-kilometer radius. The airspace environment is complex, with high-rise buildings ranging in height from 50 to 300 meters and frequent flocks of migratory birds. During peak delivery periods, up to 300 drones can be in the air simultaneously, creating a high-density distribution within the 8,000 cubic meters of available airspace, with an average of one drone per 100 cubic meters. When multiple drones perform delivery missions, they frequently encounter other drones, buildings, and flocks of birds. Traditional single-drone obstacle avoidance methods struggle to cope with such complex multi-drone collaborative obstacle avoidance requirements.

[0035] In related technologies, obstacle avoidance control for a swarm of drones can be achieved by adopting a fixed-priority obstacle avoidance method. The following describes a scenario where the flight control method for a swarm of drones in related technologies is used.

[0036] A logistics company uses a fixed-priority obstacle avoidance method to handle drone swarm deliveries. This method assigns each drone a priority level of 1-5 based on the urgency of the mission, requiring low-priority drones to completely avoid high-priority drones. When a potential collision risk is detected, the system calculates the priority differences between drones and forces low-priority drones to deviate from their original routes by at least 50 meters. However, in actual operation, when more than 10 drones face obstacle avoidance requirements simultaneously, this method causes low-priority drones in levels 3-5 to change their flight paths on average every three minutes, increasing energy consumption by more than 30% and triggering a chain reaction of obstacle avoidance. Furthermore, the method's response time to sudden dynamic obstacles, such as flocks of birds, generally exceeds 5 seconds, posing a serious safety hazard.

[0037] The multi-UAV swarm flight control method described in the embodiments of this application achieves efficient and reliable swarm obstacle avoidance control by calculating the density distribution in multiple search directions in real time, selecting the optimal obstacle avoidance direction, and utilizing a rehearsal mechanism to coordinate the obstacle avoidance space. This not only avoids the frequent obstacle avoidance and chain reaction issues associated with traditional methods, but also significantly improves the real-time and safety of obstacle avoidance. The following describes scenarios in which the multi-UAV swarm flight control method described in this application is used.

[0038] The logistics distribution system using the solution of this application evenly divides the airspace into 16 search directions, and calculates the density distribution of drones within a range of 100 meters in real time in each direction. When a drone needs to avoid obstacles, it will select the optimal obstacle avoidance direction based on the density difference within 200 milliseconds, and coordinate the obstacle avoidance space with surrounding drones through a rehearsal mechanism. For example, when a delivery drone encounters a flock of about 50 sparrows at an altitude of 150 meters, the system immediately assesses that the density value in the southeast direction is 40% lower than that in other directions, selects this direction for obstacle avoidance, and notifies other drones in this direction to reserve at least 30 cubic meters of obstacle avoidance space. In this way, the system controls the deviation of the drone's route to a minimum range, the obstacle avoidance process takes only 2.5 seconds on average, and the energy loss increases by no more than 10%. At the same time, it ensures that the safe distance from other drones is always greater than 15 meters.

[0039] It can be seen that the flight control method of a multi-UAV cluster in the embodiment of the present application can not only realize multi-UAV obstacle avoidance control, but also effectively solve the energy waste problem caused by the traditional fixed priority method, thereby realizing more intelligent and efficient UAV cluster obstacle avoidance control.

[0040] For ease of understanding, the following describes the process of the method provided by this implementation in combination with the above scenario. Figure 1 , which is a flow chart of the flight control method of a multi-UAV cluster in an embodiment of the present application.

[0041] S101: Obtain the main body position data and preset flight mission of the main body UAV, and broadcast the main body position data to the surrounding space.

[0042] The main UAV represents the entity currently executing control; the main position data refers to a data set containing the UAV's current 3D coordinate position, flight speed, heading angle, and other state information; the preset flight mission refers to a pre-planned mission instruction set containing information such as the target location, flight path, and mission priority; and the surrounding space refers to the 3D spatial area within the preset communication radius centered on the main UAV. Broadcasting refers to sending data information to all UAVs in the surrounding space via wireless communication.

[0043] When a drone swarm begins a flight mission, the control device must first obtain initial state information and establish communication with surrounding drones. Specifically, the control device uses onboard sensors to obtain GPS positioning data and inertial measurement unit data from the main drone, and combines this data with a data fusion algorithm to obtain accurate main drone position data. Simultaneously, the control device reads pre-set flight mission information from the mission planning module, including the starting position, target position, waypoint sequence, and mission constraints. The control device then broadcasts this encapsulated main drone position data via radio signals to a 360-degree radius, enabling other drones within the communication radius to receive this information.

[0044] During implementation, the solution may encounter issues such as degraded GPS signal quality and unstable communication links due to inclement weather or electromagnetic interference. To address this, the control device can employ an adaptive data fusion strategy: when GPS signal quality is poor, the IMU data is weighted more heavily, while auxiliary positioning methods such as visual odometry are utilized. When communication interference is detected, communication reliability is improved by increasing transmit power, reducing data transmission rates, and employing anti-interference coding. For example, in environments with strong electromagnetic interference, the original 1Hz position broadcast frequency can be reduced to 0.5Hz, and forward error correction coding can be employed to enhance the anti-interference capability of data transmission.

[0045] S102: Receive adjacent position data in multiple search directions, determine the number of adjacent drones, and calculate the density value of the main drone in the search direction.

[0046] Among them, the search direction represents a set of preset directions radiating outward from the main UAV, usually dividing the 360-degree space evenly into several sectors; the adjacent position data refers to the status data containing position, speed and other information received from surrounding UAVs; adjacent UAVs refer to other UAVs within a preset distance range; the main body density value represents the distribution density of UAVs in a unit space in a certain search direction, which is used to quantitatively describe the congestion state of the local space.

[0047] After completing its position broadcast, the control device needs to continuously receive and process status information from surrounding drones. Specifically, the control device first divides the 360-degree space into eight or 16 equally sized sector-shaped search areas. It then receives and analyzes position data packets sent by nearby drones in each search direction. For each search direction, it counts the number of drones within a preset distance range and, combined with the spatial volume of the search area, calculates a normalized density value. This density value reflects the density of drones in that direction, providing an important basis for subsequent obstacle avoidance decisions.

[0048] In some embodiments, density calculation can be achieved in a variety of ways: Optionally, the control device can adopt a density calculation method based on distance weighting, first dividing the search area into three sub-areas: near, middle, and far, and assigning different weight coefficients to each sub-area, then multiplying the number of drones in each sub-area by the corresponding weight and summing the results, and finally dividing by the total volume to obtain a weighted density value; Optionally, the control device can adopt a continuous density estimation method based on a Gaussian kernel function, treating each drone as a density center, calculating its density contribution to the surrounding space through a Gaussian kernel function, and finally superimposing to obtain a continuous density distribution field. It is understandable that other spatial statistical methods can also be used to calculate the density distribution, which is not limited here. In addition, in order to improve the accuracy of the density calculation, it is also necessary to consider the motion state of the drone, and speed prediction can be introduced to correct the density estimation.

[0049] During implementation, the high-speed movement of the drone may cause rapid changes in density distribution. To address this, the control device can employ a dynamic time window density estimation strategy: the time window size for density calculation is adaptively adjusted based on the drone's average speed. When the speed is high, the time window is shortened to improve timeliness, while when the speed is low, the time window is appropriately extended to improve stability. Furthermore, a Kalman filter is introduced to smooth the density values ​​and suppress the noise caused by transient fluctuations. For example, when the average speed is detected to exceed 10 m / s, the density calculation time window is shortened from 2 seconds to 1 second, and the filter's process noise covariance is increased to improve tracking response speed.

[0050] S103: Obtain proximity density values ​​of adjacent drones in multiple search directions, and determine the location of the obstacle when the main drone detects an obstacle.

[0051] Among them, the proximity density value represents the local density information of the location of the neighboring drone received from it; obstacles refer to static or dynamic objects that may affect the normal flight of the drone; obstacle position refers to the coordinate position of the obstacle in three-dimensional space and the spatial range information it occupies.

[0052] While acquiring the drone's density, the control device also collects density information from nearby drones and performs obstacle detection. Specifically, the control device receives local density data broadcast by nearby drones via wireless communication and sorts the received data according to the search direction. Simultaneously, it activates onboard multi-source sensing equipment, including lidar, millimeter-wave radar, and visual sensors, to perform a full-scale scan of the surrounding environment. When an obstacle is detected, it fuses the data from multiple sensors to calculate the obstacle's precise three-dimensional position, size, shape, and other characteristic information, and creates a grid map of the obstacle's occupancy.

[0053] In some embodiments, the determination of the obstacle position can be achieved in a variety of ways: Optionally, the control device can adopt an obstacle detection method based on multi-sensor fusion, first performing ground segmentation and clustering on the laser point cloud data to identify potential obstacles, then combining the distance and speed measurement results of the millimeter wave radar, and finally confirming the obstacle type and precise outline through visual recognition; Optionally, the control device can adopt an environmental modeling method based on probabilistic occupancy grids, dividing the space into uniform grids, calculating the occupancy probability of each grid through Bayesian updating, and introducing a time decay factor to handle dynamic obstacles. It is understandable that other obstacle detection and positioning algorithms can also be used, which are not limited here. In addition, the blind spot problem of the sensor needs to be considered, and the blind spot information can be supplemented by reasonably arranging sensors or using observation data from nearby drones.

[0054] S104: Determine an obstacle avoidance direction and an obstacle avoidance reverse direction based on the main body density value and the neighboring density value.

[0055] The obstacle avoidance direction represents the preferred direction of movement for the drone to perform obstacle avoidance maneuvers; the obstacle avoidance reverse direction represents the search direction opposite to the obstacle avoidance direction, used to coordinate multi-drone obstacle avoidance behaviors; and the density difference represents the numerical difference between the drone's density value and the adjacent density value, used to quantify the difference in spatial congestion in different directions. The preset density threshold is a reference standard for determining spatial congestion.

[0056] After obtaining complete density distribution information, the control device needs to plan an obstacle avoidance strategy based on the density analysis results. Specifically, the control device first calculates the difference between the main body density value and the corresponding adjacent density value for each search direction, forming a complete density difference distribution map. It then analyzes the density differences in all search directions and selects the direction with the largest density difference as the obstacle avoidance direction, ensuring that the obstacle avoidance process is carried out in a relatively open area. Finally, the search direction opposite to the obstacle avoidance direction is determined as the opposite direction of the obstacle avoidance, which is used for subsequent obstacle avoidance coordination. This density difference-based direction selection strategy allows the drone to naturally move towards areas with lower density, reducing the impact on other drones.

[0057] In some embodiments, the obstacle avoidance direction can be determined in a variety of ways: Optionally, the control device can adopt a direction decision method based on fuzzy logic, taking multiple factors such as density difference, obstacle distance, task constraints, etc. as input variables, and calculating the fitness score of each direction through the designed fuzzy rule set, and selecting the direction with the highest score as the obstacle avoidance direction; Optionally, the control device can adopt a direction planning method based on artificial potential field, converting the density distribution into a repulsive field, the obstacle generates additional repulsive force, the task target generates gravitational force, and the obstacle avoidance direction is determined by the direction of the resultant force. It is understandable that other decision algorithms can also be used to determine the obstacle avoidance direction, which is not limited here. In addition, the smooth transition problem of the obstacle avoidance direction needs to be considered, and a direction filter can be introduced to avoid drastic direction changes.

[0058] During implementation, multiple directions may have similar density differences, making it difficult to determine the optimal obstacle avoidance direction. To address this, the control device can employ a multi-criteria decision-making approach. In addition to density differences, it also considers factors such as the deviation angle from the current flight direction, the change in distance from the target point, and energy consumption. By designing a comprehensive evaluation function, each candidate direction is assigned a comprehensive score.

[0059] S105: Obtain an obstacle avoidance rehearsal space of an adjacent UAV in the opposite direction of the obstacle avoidance direction.

[0060] Among them, the obstacle avoidance rehearsal space represents the three-dimensional spatial range that can be used to perform obstacle avoidance maneuvers; the obstacle avoidance collaboration data contains information such as the current status, expected trajectory, and obstacle avoidance intention of nearby drones; the available obstacle avoidance space refers to the spatial area that can actually be used for obstacle avoidance after considering all constraints; and the maneuverability performance constraints include physical limitations of the aircraft such as maximum acceleration and steering angular velocity.

[0061] After determining the obstacle avoidance direction, the control device needs to assess the availability of the obstacle avoidance space. Specifically, the control device first sends an obstacle avoidance rehearsal request to a neighboring drone in the opposite direction of the obstacle avoidance direction. The request contains information such as the drone's position, speed, and obstacle avoidance intent. After receiving the obstacle avoidance coordination data returned by the neighboring drone, it combines the obstacle position information to construct a four-dimensional space-time obstacle avoidance rehearsal model. By considering the kinematic constraints and dynamic characteristics of each drone, the control device calculates the available obstacle avoidance space that meets the safety spacing requirements. This rehearsal-based spatial assessment method can identify potential conflict risks in advance and ensure the safety of obstacle avoidance actions.

[0062] In some embodiments, the acquisition of the obstacle avoidance rehearsal space can be achieved in a variety of ways: Optionally, the control device can use a spatial rehearsal method based on a probability map to convert the expected trajectory of each drone into a time-varying probability occupancy map, calculate the time-varying free space distribution through probability superposition, and finally determine the obstacle avoidance space that meets the safety requirements; Optionally, the control device can use dynamic spatial planning based on the speed barrier method to convert the movement of other drones and obstacles into a constrained area in the speed space, and determine the safe obstacle avoidance space by solving the feasible speed set. It is understandable that other spatial planning algorithms can also be used for obstacle avoidance rehearsal, which is not limited here. In addition, the prediction error caused by communication delay needs to be considered, and this uncertainty can be compensated by increasing the safety margin.

[0063] During implementation, multiple drones may simultaneously request obstacle avoidance rehearsals, leading to a surge in computational load. To address this, the control system can employ a hierarchical rehearsal strategy: Different rehearsal priorities are assigned based on the urgency of the obstacle avoidance process. Accurate trajectory rehearsals are performed for more urgent requests, while simplified space reservations are used for less urgent requests. A rehearsal resource management mechanism is also implemented to limit the number of simultaneous rehearsals.

[0064] S106: Determine an obstacle avoidance adjustment path based on the obstacle avoidance rehearsal space, obstacle location, obstacle avoidance direction, and preset flight mission, and calculate the flight path of the main UAV.

[0065] Among them, the obstacle avoidance adjustment path represents the local path segment that temporarily deviates from the original route to avoid obstacles; the flight path refers to the complete flight trajectory after taking into account the obstacle avoidance needs; the preset flight mission includes mission target points, time constraints, energy consumption limits and other mission requirements; the path optimization indicators include evaluation parameters such as path length, energy consumption, and mission deviation.

[0066] After acquiring the obstacle avoidance rehearsal space, the control device needs to plan a specific obstacle avoidance path and integrate it into the overall flight path. Specifically, the control device first constructs a three-dimensional path search graph within the obstacle avoidance rehearsal space, setting key path points including the starting point, obstacle avoidance point, and re-entry point. Then, based on a multi-objective optimization criterion, the control device generates an obstacle avoidance adjustment path within the search graph that satisfies dynamic constraints. The obstacle avoidance adjustment path is then smoothly connected to the original flight mission path to ensure a continuous and flyable path transition. Finally, the feasibility of the overall flight path is evaluated, including checking whether the path meets conditions such as safety spacing requirements, energy constraints, and mission time limits.

[0067] In some embodiments, path planning can be achieved in a variety of ways: Optionally, the control device can use a fast path planning method based on the improved RRT* algorithm, randomly sample and construct a path tree in the obstacle avoidance preview space, adjust the growth direction of the tree through heuristic optimization criteria, and optimize and reconnect the path considering dynamic constraints; Optionally, the control device can use an online path planning method based on model predictive control to establish a UAV motion model, and generate a smooth obstacle avoidance path that meets multiple constraints by solving the optimal control problem in a finite time domain. It is understandable that other path planning algorithms can also be used, which are not limited here. In addition, it is necessary to consider the problem of computational efficiency. A hierarchical planning strategy can be used to quickly generate a rough path first, and then perform local fine optimization.

[0068] During implementation, there may be significant conflicts between the obstacle avoidance path and the mission objectives. To address this, the control device can employ a multi-objective optimization strategy with adaptive weights. This strategy involves designing a comprehensive evaluation function that incorporates multiple optimization objectives, such as safety, mission efficiency, and energy consumption, and dynamically adjusting the weights of each objective based on the current situation. For example, when the distance to an obstacle is close, the safety weighting is increased; when the risk of mission delay increases, the mission efficiency weighting is increased.

[0069] S107: Generate flight control instructions for the main UAV based on the flight path to control the main UAV to perform a preset flight mission.

[0070] Among them, the flight control instruction represents the underlying control quantity used to control the drone to perform specific flight actions; the execution cycle represents the update time interval of the control instruction; the control quantity includes parameters such as thrust, attitude angle, and angular velocity; the feedback correction value refers to the correction value of the control instruction based on the actual execution effect.

[0071] After completing path planning, the control device needs to convert the high-level path into actual, executable control instructions. Specifically, the control device first discretizes the planned flight path into a series of desired state points along a time series. Then, based on the drone's dynamic model, it calculates the specific control variables required to reach each state point. These control variables are then packaged into a sequence of control instructions according to the execution cycle. During execution, sensor feedback is used to monitor the execution results in real time, dynamically adjusting control parameters to ensure the drone accurately tracks the planned path. This closed-loop control approach effectively compensates for external interference and model errors.

[0072] In some embodiments, the generation of control instructions can be achieved in a variety of ways: Optionally, the control device can adopt a nonlinear controller design method based on the backstepping method, decompose the path tracking problem into two subsystems of attitude control and position control, construct a stable control law by designing the Lyapunov function layer by layer, and finally generate specific control instructions; Optionally, the control device can adopt an intelligent control method based on adaptive dynamic programming, establish a state-control mapping relationship through online learning, achieve rapid adaptation to environmental changes, and dynamically optimize the control strategy. It is understandable that other controller design methods can also be used, which are not limited here. In addition, the smoothness requirements of the control instructions need to be considered, and drastic control inputs can be avoided by adding control quantity change rate constraints.

[0073] During implementation, there may be significant deviations between actual and expected results due to external factors such as wind disturbances. To address this, the control device can employ a layered robust control strategy: In the outer layer, an adaptive observer estimates the magnitude and direction of external disturbances and compensates for them in real time; in the inner layer, robust control methods such as sliding mode control are employed to ensure system stability in the presence of disturbances.

[0074] The following is a more detailed description of the process of the method provided by this implementation. Figure 2 , is another flow chart of the flight control method for a multi-UAV cluster in an embodiment of the present application.

[0075] S201: Obtain the main body position data and preset flight mission of the main body UAV, and broadcast the main body position data to the surrounding space.

[0076] Referring to step S101 , the control device broadcasts the body position data to the surrounding space.

[0077] S202: Receive adjacent position data in multiple search directions, determine the number of adjacent drones, and calculate the density value of the drone in the search direction.

[0078] Referring to step S102 , the control device determines the body density value.

[0079] In some embodiments, the control device determines the surrounding detection range based on environmental factors, that is, the control device obtains environmental parameters in multiple search directions; the environmental parameters include wind speed, wind direction, air pressure and temperature; weights are assigned to multiple search directions based on the environmental parameters to obtain search direction weight coefficients; the effective detection distances in multiple search directions are adjusted according to the search direction weight coefficients, and a preset distance range is determined based on the effective detection distance.

[0080] Among them, environmental parameters represent the set of external environmental factors that affect the flight of the drone; wind speed represents the rate of air flow; wind direction refers to the direction angle of air flow; air pressure represents the numerical value of atmospheric pressure; temperature refers to the temperature value of the ambient air; search direction weight coefficient is used to indicate the importance of different search directions; effective detection distance represents the effective perception range of the sensor under specific environmental conditions; preset distance range refers to the actual detection range after considering environmental influences.

[0081] Before executing a search mission, the control device must fully consider the impact of environmental factors on detection performance. Specifically, the control device first uses onboard sensors to obtain environmental parameters such as wind speed, wind direction, air pressure, and temperature in each search direction. It then analyzes the impact of these parameters on detection performance and establishes an environment-performance mapping relationship. A weight coefficient is then calculated for each search direction based on this mapping relationship, with a larger weight indicating better detection conditions in that direction. The nominal detection range is then corrected based on the weight coefficient to obtain the actual effective detection range. Finally, based on the effective detection range, a preset distance range is determined for subsequent density calculations and obstacle avoidance decisions. This detection range adjustment mechanism, based on environmental adaptation, can improve detection reliability.

[0082] In some embodiments, the processing and weight calculation of environmental parameters can be achieved in a variety of ways: Optionally, the control device can adopt a weight calculation method based on multi-factor evaluation, first construct an environmental factor evaluation matrix, standardize parameters such as wind speed, wind direction, air pressure, temperature, etc., and then determine the weight of each factor through the hierarchical analysis method, and finally calculate the comprehensive score of each search direction; Optionally, the control device can adopt a performance prediction method based on a physical model to establish a sensor performance model that takes environmental factors into account, and obtain the detection performance attenuation law under different conditions through theoretical calculation, and then determine the direction weight and effective detection distance. It is understandable that other environmental adaptation algorithms can also be used, which are not limited here. In addition, the time-varying characteristics of environmental parameters need to be considered, and real-time adjustment of weights can be achieved by introducing a dynamic update mechanism.

[0083] S203: Obtain proximity density values ​​of adjacent drones in multiple search directions, and determine the location of the obstacle when the main drone detects an obstacle.

[0084] Referring to step S103 , the control device determines the location of the obstacle.

[0085] S204: Determine the obstacle avoidance direction and the obstacle avoidance reverse direction based on the main body density value and the neighboring density value.

[0086] Referring to step S104 , the control device determines the obstacle avoidance direction and the obstacle avoidance reverse direction.

[0087] In some embodiments, the control device selects a direction based on the density difference, that is, the control device calculates the density difference in multiple search directions based on the main body density value and the adjacent density value; the density difference is the difference between the main body density value and the adjacent density value in the corresponding search direction; the search direction with the largest density difference is selected as the obstacle avoidance direction, and the search direction opposite to the obstacle avoidance direction is determined as the obstacle avoidance reverse direction.

[0088] Among them, the density difference is a quantitative indicator describing the uneven distribution of drones in different search directions; the search direction refers to the preset angle direction radiating outward from the main drone; the main density value indicates the density of drones in the current search direction; the adjacent density value refers to the local density information of the location of adjacent drones; the obstacle avoidance direction is used to indicate the preferred direction of the drone's obstacle avoidance movement; the obstacle avoidance reverse direction indicates the search direction opposite to the obstacle avoidance direction, which is used to coordinate the obstacle avoidance behavior of multiple drones; the weight coefficient is used to indicate the degree of influence of different factors on the direction selection.

[0089] After obtaining complete density distribution information, the control device needs to determine the optimal obstacle avoidance direction through density difference analysis. Specifically, the control device first traverses all search directions and calculates the numerical difference between the main body density value and the corresponding adjacent density value in each direction. The calculated density difference is then normalized to eliminate the influence of density magnitude in different directions. The normalized density difference values ​​of all search directions are then compared, and the direction with the largest difference is selected as the obstacle avoidance direction, indicating movement towards a relatively open area. Finally, the search direction opposite to the selected obstacle avoidance direction is determined as the obstacle avoidance reverse direction, which is used for subsequent obstacle avoidance coordination. This density difference-based direction selection strategy allows the drone swarm to naturally tend towards a uniform distribution.

[0090] In some embodiments, direction selection based on density difference can be achieved in a variety of ways: Optionally, the control device can adopt a decision-making method based on fuzzy logic, first taking factors such as density difference, obstacle distance, and task deviation as input variables for fuzzy processing, and then calculating the comprehensive score of each direction through the designed fuzzy rule set, and finally obtaining the final obstacle avoidance direction through the center of gravity method; Optionally, the control device can adopt a direction planning method based on artificial potential field, converting the density difference into a repulsive potential field, constructing a synthetic potential field containing task gravity and obstacle repulsion, and determining the obstacle avoidance direction through the potential field gradient. It is understandable that other decision-making algorithms can also be used to achieve direction selection, which is not limited here. In addition, the smoothness requirement of direction selection needs to be considered, and drastic direction switching can be avoided by adding direction change rate constraints.

[0091] S205: Send an obstacle avoidance rehearsal request to a neighboring UAV in the opposite direction of the obstacle avoidance direction.

[0092] Among them, the obstacle avoidance rehearsal request represents the data packet that initiates the rehearsal collaboration; the obstacle avoidance reverse direction represents the search direction opposite to the obstacle avoidance direction; the adjacent drones refer to other drones located in the obstacle avoidance reverse direction; the rehearsal collaboration data includes information such as the rehearsal start time, rehearsal duration, and rehearsal space range; the response data includes information such as the current status and expected trajectory of the receiving drone.

[0093] After determining the obstacle avoidance direction, the control device needs to implement multi-drone obstacle avoidance coordination by sending a rehearsal request. Specifically, the control device first determines the opposite direction of the obstacle avoidance direction based on the obstacle avoidance direction and identifies nearby drones in that direction. It then constructs a rehearsal request packet containing information such as the rehearsal time window and spatial range. This request packet is then sent to the relevant drones via a wireless communication channel. The device then waits for and receives responses from nearby drones. Finally, the received responses are acknowledged to establish a rehearsal coordination relationship. This request-response-based rehearsal coordination mechanism ensures information synchronization and behavioral coordination during the obstacle avoidance process.

[0094] In some embodiments, the sending and processing of rehearsal requests can be achieved in a variety of ways: Optionally, the control device can adopt a time-slice-based collaborative communication method, first dividing the rehearsal process into multiple time segments, then sending requests in order of priority within each time segment and waiting for responses, then retransmitting timed requests, and finally confirming the establishment of a collaborative relationship through a handshake mechanism; Optionally, the control device can adopt a state machine-based rehearsal management method, establish a rehearsal state machine including idle, waiting, collaborative and other states, and achieve state transitions through event triggering to ensure the orderly progress of the rehearsal process. It is understandable that other collaborative communication methods can also be used, which are not limited here. In addition, communication delays and packet loss issues need to be considered, and communication reliability can be improved by setting a retransmission mechanism and timeout processing.

[0095] During the implementation of the solution, you may encounter the problem of multiple drones sending rehearsal requests simultaneously, causing communication conflicts. To address this, the control device can adopt a distributed conflict handling strategy: introducing a timestamp-based request arbitration mechanism to coordinate concurrent rehearsal requests. Specifically, the following processing flow can be designed: Request_Priority = f(timestamp, distance, urgency), where timestamp is the timestamp of the request initiation, distance is the distance to the obstacle avoidance area, and urgency is the urgency of obstacle avoidance. In this way, the system can process rehearsal requests in order of comprehensive priority to avoid communication congestion. At the same time, a request queue management mechanism is established to automatically adjust the request sending frequency and retransmission interval when the communication load is detected to be too high. For example, when more than 10 rehearsal requests are received within 1 second, the system will automatically extend the retransmission wait time and give priority to high-priority requests.

[0096] S206: Receive obstacle avoidance cooperation data returned by the neighboring UAV in the opposite direction of obstacle avoidance.

[0097] Among them, obstacle avoidance collaborative data represents the data set used for collaborative obstacle avoidance decision-making; data timestamp refers to the precise moment when the data is generated, which is used to judge the timeliness of the data; data credibility represents the evaluation result of the data accuracy; response time limit refers to the maximum waiting time expected to receive collaborative data; data integrity is used to indicate the completeness of the received data; data priority represents the importance ranking of different types of collaborative data.

[0098] After sending an obstacle avoidance rehearsal request, the control device needs to wait for and process the collaborative data returned by the neighboring drone. Specifically, the control device first establishes a data reception buffer and sets a response waiting timeout. It then performs a timestamp check and integrity verification on the received collaborative data. It then parses the data packets for information such as flight status, predicted trajectory, and obstacle avoidance intent. Finally, it filters and prioritizes the data based on its timeliness and credibility, eliminating invalid or expired data. This collaborative data processing mechanism ensures the reliability of subsequent obstacle avoidance decisions.

[0099] In some embodiments, the reception and processing of collaborative data can be achieved in a variety of ways: Optionally, the control device can adopt a data reception method based on a reliable transmission protocol, and by setting a data verification and confirmation mechanism, automatically request retransmission when a data packet is detected to be lost or damaged, and use sliding window technology to achieve orderly reception and cache management of data; Optionally, the control device can adopt a data screening method based on trust evaluation, establish a trust model for each nearby drone, and dynamically adjust the data credibility threshold by comprehensively considering factors such as historical data quality and communication stability. It is understandable that other data reception and processing methods can also be used, which are not limited here. In addition, it is necessary to consider the communication bandwidth limitation, and the transmission load can be reduced through data compression and importance sampling.

[0100] During implementation, conflicts may arise between collaborative data returned by multiple nearby drones. To address this, the control device can employ a data fusion strategy based on evidence theory: First, a basic trust probability is assigned to each data source. Then, the Dempster-Shafer evidence theory is used to fuse the data, resolving data uncertainty and conflicts.

[0101] S207: Obtain an obstacle avoidance rehearsal space by calculation based on the obstacle position and the obstacle avoidance coordination data.

[0102] Among them, the obstacle avoidance rehearsal space represents a four-dimensional space-time region, which includes three-dimensional spatial coordinates and time dimensions; the safety margin refers to the additional safety distance reserved; the spatial resolution refers to the minimum unit size of the space division; the time resolution refers to the minimum time step of the rehearsal simulation; the reachability index is used to evaluate whether the spatial point can be reached safely; the constraints include the kinematic constraints and dynamic constraints of the UAV.

[0103] After acquiring the collaborative obstacle avoidance data, the control device must synthesize various pieces of information to construct a rehearsal space. Specifically, the control device first maps the obstacle positions and trajectory information from the collaborative data into a unified spatiotemporal coordinate system. It then considers the motion constraints of each drone and calculates the reachable spatial range at different times. The spatiotemporal space is then discretized into a grid structure, and the occupancy status and reachability index are calculated for each grid. Finally, a safety margin is added to generate a representation of the obstacle avoidance rehearsal space that takes all constraints into account. This rehearsal space construction method provides reliable spatial constraints for subsequent path planning.

[0104] In some embodiments, the construction of the obstacle avoidance rehearsal space can be achieved in a variety of ways: Optionally, the control device can use a spatial representation method based on a probability flow field to convert the motion trajectory of each drone into a time-varying probability density distribution, and obtain the available space area through probability superposition and threshold processing, while considering the influence of uncertainty propagation; Optionally, the control device can use a modeling method based on a spatiotemporal occupancy grid to construct a four-dimensional grid map, and calculate the accessibility and risk of each grid through a dynamic programming algorithm to achieve accurate modeling of complex dynamic environments. It is understandable that other spatial modeling methods can also be used, which are not limited here. In addition, it is necessary to consider the issue of computational efficiency, and a multi-resolution representation can be used to refine the key areas.

[0105] During implementation, the computational load during the rehearsal space construction process may be excessive. To address this, the control device can employ a hierarchical and progressive space construction strategy: first, a global rehearsal space is quickly constructed at a coarse resolution to obtain a rough estimate of the feasible region; then, the potential obstacle avoidance path region is locally refined to improve the spatial resolution of key areas; and finally, parallel computing technology is used to accelerate the processing.

[0106] In some embodiments, the control device will perform collaborative obstacle avoidance, that is, the control device will determine the current flight status and expected flight trajectory of the adjacent drone in the opposite direction of obstacle avoidance based on the obstacle avoidance collaborative data; calculate the available obstacle avoidance space of the main drone in the direction of travel based on the current flight status, expected flight trajectory and obstacle position; determine the obstacle avoidance rehearsal space based on the available obstacle avoidance space and the maneuverability constraints of the main drone.

[0107] Among them, obstacle avoidance collaborative data represents the information set used for multi-machine collaborative obstacle avoidance; the current flight state includes kinematic parameters such as position, speed, and acceleration; the expected flight trajectory refers to the predicted motion path of the nearby drone in the future; the direction of travel represents the expected motion direction of the main drone; the available obstacle avoidance space refers to the optional maneuvering space that meets safety requirements; the maneuverability constraints include physical limitations such as maximum speed, acceleration, and steering angular velocity; the obstacle avoidance rehearsal space represents the actual available space range after considering all constraints.

[0108] After acquiring the collaborative obstacle avoidance data, the control device needs to conduct a comprehensive analysis and construct a usable obstacle avoidance rehearsal space. Specifically, the control device first analyzes the state information in the collaborative obstacle avoidance data to extract key parameters such as the position and speed of the adjacent drone. It then uses a prediction algorithm to deduce the future trajectory of the adjacent drone based on the current state, while also considering the uncertainty range of the prediction. The obstacle position information is then spatially superimposed with the predicted trajectory to determine the spatial occupancy at each moment. The reachable range at each moment is then calculated based on the direction and maneuverability of the main drone. Finally, an intersection operation is performed to obtain an obstacle avoidance rehearsal space that satisfies all constraints. This rehearsal space construction method provides reliable spatial constraints for subsequent path planning.

[0109] In some embodiments, the construction of the obstacle avoidance rehearsal space can be achieved in a variety of ways: Optionally, the control device can adopt a spatial representation method based on a probability map, first converting the predicted trajectory of the adjacent drone into a time-varying probability density distribution, then superimposing the spatial occupancy probability of the obstacle, then determining the boundary of the available space by setting a probability threshold, and finally performing pruning optimization considering the maneuverability constraint; Optionally, the control device can adopt a modeling method based on convex space decomposition, first decomposing the complex environment into a series of simple convex areas, then calculating the accessibility index within each convex area, then connecting the feasible convex areas through a dynamic programming algorithm, and finally generating a complete rehearsal space representation. It is understandable that other spatial modeling methods can also be used, which are not limited here. In addition, it is necessary to consider the issue of computational efficiency, and a multi-resolution representation can be used to refine the key areas.

[0110] During the implementation of the plan, you may encounter inaccurate preview space due to prediction errors. To address this, the control device can adopt an adaptive uncertainty handling strategy: introducing a spatial expansion mechanism based on prediction credibility. The larger the prediction error, the more conservative the corresponding spatial constraints. Specifically, a spatial expansion function can be designed: Expansion(t) = base_size * (1 + β * uncertainty(t)), where t is the prediction time, base_size is the basic safety distance, β is the expansion coefficient, and uncertainty(t) is the prediction uncertainty function. In this way, while ensuring safety, overly conservative spatial constraints are avoided. At the same time, a dynamic update mechanism for the preview space is established. When it is detected that the prediction error exceeds the threshold, the reconstruction process is triggered in a timely manner.

[0111] S208: Determine an obstacle avoidance adjustment path based on the obstacle avoidance rehearsal space, obstacle location, obstacle avoidance direction, and preset flight mission, and calculate the flight path of the main UAV.

[0112] Referring to step S106 , the control device calculates the flight path.

[0113] S209: Detect dynamic obstacles on the flight path and calculate a preset trajectory of the dynamic obstacles.

[0114] Among them, dynamic obstacles refer to obstacle targets with motion characteristics in space, such as other aircraft, moving objects, etc.; preset trajectory refers to the predicted result of the motion path of dynamic obstacles in the future; obstacle state quantities include kinematic parameters such as position, velocity, and acceleration; motion mode refers to the typical motion characteristics of obstacles, such as uniform linear motion, turning motion, etc.; prediction time domain refers to the time span of trajectory prediction; prediction uncertainty is used to describe the credibility of the trajectory prediction results.

[0115] After completing the initial flight path planning, the control device needs to continuously monitor for dynamic obstacles that may appear along the path. Specifically, the control device first detects dynamic obstacles in real time through multi-source sensor fusion, obtaining their current position and motion state. It then identifies the obstacle's motion pattern and behavioral characteristics based on historical observation data. A prediction model is then established, combining kinematic constraints and environmental information to calculate the obstacle's possible trajectory within the predicted time domain. The uncertainty of the prediction results is also evaluated, generating trajectory prediction clusters with confidence intervals. This dynamic prediction mechanism can detect potential collision risks in advance, providing a basis for path updates.

[0116] In some embodiments, dynamic obstacle trajectory prediction can be achieved through a variety of methods: Optionally, the control device can adopt a trajectory prediction method based on interactive multiple models. First, a model library containing multiple typical motion modes, such as constant speed models, steering models, acceleration models, etc., is constructed. Then, the state of each model is estimated through a filter bank. The model is adaptively switched based on the model probability, and finally the predicted trajectory of the obstacle is obtained by fusion. Optionally, the control device can adopt a trajectory generation method based on deep learning, using a long short-term memory network (LSTM) to establish a sequence prediction model. The model inputs the historical trajectory sequence and environmental characteristics, outputs the state prediction for the future moment, and improves the prediction robustness through ensemble learning. It is understood that other trajectory prediction algorithms can also be used to achieve dynamic obstacle motion prediction, which is not limited here. In addition, considering that the prediction accuracy gradually decreases over time, an adaptive prediction time domain mechanism can be introduced to dynamically adjust the prediction range according to the obstacle motion characteristics.

[0117] During implementation, dynamic obstacles can cause significant deviations in predictions due to sudden changes in their motion patterns. To address this, the control system can employ a hierarchical and progressive prediction correction strategy. At the prediction level, a multi-hypothesis prediction tree is constructed to maintain multiple possible motion hypotheses. When a mode shift is detected, the system quickly switches to the most suitable prediction branch. At the evaluation level, an anomaly detection mechanism based on the Mahalanobis distance is introduced to calculate the deviation between the observed and predicted values ​​in real time. When the deviation exceeds a threshold, a reconstruction of the prediction model is triggered.

[0118] S210: Update the obstacle avoidance adjustment path and the flight path based on the preset trajectory.

[0119] Among them, the obstacle avoidance adjustment path represents the local path segment replanned to avoid collision with dynamic obstacles; the path update trigger condition refers to the judgment basis for the need for path adjustment; the collision risk degree represents the estimated value of the possibility of collision with dynamic obstacles; the path smoothness refers to the continuity and flyability indicators of the path; the transition interval represents the spatial range of switching from the original path to the new path; the real-time requirement refers to the time constraint that the path update must meet.

[0120] After obtaining the preset trajectory of a dynamic obstacle, the control device needs to promptly assess the collision risk and update the flight path. Specifically, the control device first calculates the spatiotemporal collision risk based on the preset trajectory of the dynamic obstacle and determines whether a path update needs to be triggered. When the collision risk exceeds a threshold, the control device replans a local obstacle avoidance path between the current position and the target point. The newly planned obstacle avoidance path is then smoothly connected to the original path to ensure that the path update does not cause drastic maneuvering changes. Finally, the overall flight path is updated and the new path tracking control is immediately executed. This dynamic path update mechanism enables real-time response to dynamic obstacles.

[0121] In some embodiments, dynamic path updates can be achieved in a variety of ways: Optionally, the control device can adopt a path replanning method based on a spatiotemporal window, first determine the feasible spatiotemporal window range according to the current state and dynamic constraints, then build a fast expansion tree within the window, use a heuristic search algorithm to find the optimal obstacle avoidance path, and finally smooth the path through a Bezier curve; Optionally, the control device can adopt an online optimization method based on model prediction, establish a prediction model that takes into account dynamic obstacle constraints, continuously update the control sequence through rolling time domain optimization, and generate an obstacle avoidance trajectory that meets multiple constraints in real time. It is understandable that other path update algorithms can also be used, which are not limited here. In addition, the problem of computational delay needs to be considered, and the response speed can be improved by pre-computing multiple alternatives in parallel.

[0122] During implementation, the rapid movement of dynamic obstacles may lead to frequent path updates. To address this, the control device can employ an adaptive path update strategy: a path stability assessment mechanism is introduced that dynamically adjusts the update frequency and range by comprehensively considering factors such as the rate of change in collision risk, the magnitude of path adjustments, and energy consumption. For example, a path update can be triggered only when the score exceeds a threshold and the minimum time interval since the last update is met. A graded response mechanism is also established, employing small, local adjustments for low-risk situations and large-scale replanning for high-risk situations. This approach ensures obstacle avoidance safety while avoiding unnecessary and frequent path adjustments.

[0123] In some embodiments, the control device will perform dynamic obstacle judgment, that is, the control device will construct a dynamic obstacle avoidance area based on the preset trajectory of the dynamic obstacle; calculate the minimum safe distance between the main UAV and the dynamic obstacle avoidance area, and determine the dynamic obstacle avoidance priority based on the minimum safe distance; adjust the obstacle avoidance strategy according to the dynamic obstacle avoidance priority and the real-time status of the obstacle avoidance rehearsal space; update the obstacle avoidance adjustment path based on the obstacle avoidance strategy, and replan the flight path in combination with the preset flight mission.

[0124] Among them, the dynamic obstacle avoidance area represents the influence range of dynamic obstacles in the time and space dimensions; the preset trajectory refers to the predicted movement path of the dynamic obstacle in the future period of time; the minimum safety distance represents the minimum interval that needs to be maintained between the drone and the dynamic obstacle avoidance area; the dynamic obstacle avoidance priority is used to indicate the urgency of the obstacle avoidance task; the obstacle avoidance strategy refers to the specific obstacle avoidance execution plan; the obstacle avoidance adjustment path represents the local path segment modified to avoid dynamic obstacles; the real-time status includes the current space occupancy and constraints.

[0125] After detecting a dynamic obstacle, the control device needs to promptly adjust the obstacle avoidance strategy and update the flight path. Specifically, the control device first constructs a time-varying obstacle avoidance area that takes into account the safety margin based on the preset trajectory of the dynamic obstacle; then calculates the shortest distance from the current position of the main UAV to the boundary of the obstacle avoidance area, and determines the obstacle avoidance priority based on the relative speed and distance; then, based on the priority and the real-time status of the obstacle avoidance rehearsal space, it selects an appropriate obstacle avoidance strategy, such as detouring, deceleration, or climbing; then, based on the selected strategy, it updates the local obstacle avoidance path and smoothly connects it to the original flight path; finally, the feasibility of the newly generated complete flight path is verified to ensure that all constraints are met. This dynamic obstacle avoidance mechanism can achieve real-time response to moving obstacles.

[0126] In some embodiments, the generation of dynamic obstacle avoidance strategies and path updates can be achieved in a variety of ways: Optionally, the control device can adopt an obstacle avoidance decision method based on threat assessment, first construct a spatiotemporal threat map, and calculate the collision risk by comprehensively considering factors such as distance, speed, and acceleration. Then, an obstacle avoidance strategy is selected based on the risk level, and then a new obstacle avoidance path is generated using a fast replanning algorithm. Finally, the continuity of the path is ensured by trajectory smoothing technology. Optionally, the control device can adopt a path optimization method based on model prediction, establish a prediction model that considers dynamic constraints, and obtain the optimal obstacle avoidance trajectory by solving the rolling time domain optimization problem, while introducing collision risk constraints to ensure safety. It is understandable that other obstacle avoidance decision and path planning algorithms can also be used, which are not limited here. In addition, the real-time computing requirements need to be considered, and the response speed can be improved through technologies such as parallel computing and plan libraries.

[0127] S211. Generate flight control instructions for the main UAV based on the flight path to control the main UAV to perform a preset flight mission.

[0128] Among them, flight control instructions represent the underlying instruction set that controls the drone to perform specific flight actions; the control cycle represents the minimum execution time interval of the instructions; the control quantity includes basic control parameters such as thrust, attitude angle, and angular velocity; the control accuracy instruction represents the error range of the expected control effect; the instruction priority is used to distinguish the execution order of different control instructions; the control constraints include physical limitations such as actuator saturation limit and dynamic constraint.

[0129] After obtaining the final flight path, the control device needs to convert the high-level path plan into a sequence of executable control instructions. Specifically, the control device first discretizes the flight path into a series of desired state points according to the control cycle. Then, based on the UAV dynamics model, it calculates the specific control variables required to reach each state point. The control variables are then constrained and smoothed to ensure they meet actuator constraints. The processed control variables are then packaged into a standard control instruction format. Finally, through a real-time feedback control loop, the control parameters are dynamically adjusted to compensate for external disturbances. This hierarchical control structure enables precise conversion of paths into control instructions.

[0130] In some embodiments, control instruction generation can be achieved in a variety of ways: Optionally, the control device can adopt a nonlinear controller based on backstepping design, first decomposing the control problem into a position control outer loop and an attitude control inner loop, then constructing a stable control law by designing Lyapunov functions layer by layer, then adaptively adjusting the control gain, and finally generating specific instructions for each actuator through the control allocation matrix; Optionally, the control device can adopt an optimal control method based on model prediction, establish a prediction model that takes all constraints into account, obtain the optimal control sequence by solving a finite time domain optimization problem, and introduce a disturbance observer for online compensation. It is understandable that other controller design methods can also be used, which are not limited here. In addition, the real-time requirements of the control instructions need to be considered, and the computational efficiency can be improved by simplifying the model and parallel computing.

[0131] During implementation, external factors such as wind disturbances may cause the control effect to deviate significantly from expectations. To address this, the control device can employ a multi-layer robust control strategy: an adaptive observer is introduced in the outer layer to estimate external disturbances in real time, sliding mode control is used in the middle layer to ensure system stability, and feedback linearization is used in the inner layer to achieve precise tracking.

[0132] In the embodiment of the present application, since a multi-machine collaborative obstacle avoidance method based on density perception is proposed and combined with a rehearsal mechanism and an environmental adaptive strategy, the system can calculate the airspace density distribution in real time, intelligently select the optimal obstacle avoidance direction, and ensure the availability of obstacle avoidance space through rehearsal coordination. At the same time, it can also dynamically adjust the detection range and obstacle avoidance strategy according to environmental parameters, effectively solving the problems of frequent obstacle avoidance, chain reaction, energy waste, etc. in traditional fixed-priority obstacle avoidance methods, thereby realizing more intelligent, efficient, safe and reliable drone cluster obstacle avoidance control.

[0133] The following describes the control device in the embodiment of the present invention from the perspective of hardware processing. Figure 3 , is a schematic diagram of a physical device structure of a control device in an embodiment of the present application.

[0134] It should be noted that Figure 3 The structure of the control device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0135] like Figure 3 As shown, the control device includes a CPU 301, which can perform various appropriate actions and processes according to the program stored in the ROM 302 or the program loaded from the storage unit 308 into the RAM 303, such as executing the method described in the above embodiment. The RAM 303 also stores various programs and data required for system operation. The CPU 301, ROM 302, and RAM 303 are connected to each other via a bus 304. An I / O interface 305 is also connected to the bus 304.

[0136] The following components are connected to the I / O interface 305: an input section 306 including an audio input device, push button switches, and the like; an output section 307 including a liquid crystal display (LCD), an audio output device, indicator lights, and the like; a storage section 308 including a hard disk and the like; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. Removable media 311, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 310 as needed, so that computer programs read from the removable media can be installed in the storage section 308 as needed.

[0137] In particular, according to an embodiment of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, an embodiment of the present invention includes a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 309 and / or installed from the removable medium 311. When the computer program is executed by the CPU 301, the various functions defined in the present invention are performed.

[0138] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram can represent a module, program segment, or part of the code, and the above-mentioned module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings.

[0139] Specifically, the control device of this embodiment includes a processor and a memory, and a computer program is stored in the memory. When the computer program is executed by the processor, the flight control method of the multi-UAV cluster provided in the above embodiment is implemented.

[0140] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the control device described in the above embodiments, or may exist independently and not be incorporated into the control device. The storage medium carries one or more computer programs, which, when executed by a processor of the control device, enable the control device to implement the flight control method for a multi-UAV swarm provided in the above embodiments.

[0141] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0142] As used in the above embodiments, the term “when” may be interpreted to mean “if” or “after” or “in response to determining that” or “in response to detecting that”, depending on the context. Similarly, the phrases “upon determining that” or “if (stated condition or event) is detected” may be interpreted to mean “if determining that” or “in response to determining that” or “upon detecting (stated condition or event)” or “in response to detecting (stated condition or event)”, depending on the context.

Claims

1. A flight control method for a multi-UAV swarm, characterized in that: Applied to a control device, the control device being provided on each drone in the multi-drone cluster, the method comprising: Obtaining the main body position data and preset flight mission of the main body drone, and broadcasting the main body position data to the surrounding space; Receive adjacent position data in multiple search directions, determine the number of adjacent drones, and calculate the density of the main drone in the search direction; the adjacent drones are drones within a preset distance range from the main drone in the search direction; Obtaining proximity density values ​​of the positions of the adjacent drones in the multiple search directions, and determining the position of the obstacle when the main drone detects an obstacle; Determining an obstacle avoidance direction and an obstacle avoidance reverse direction based on the body density value and the neighboring density value; the obstacle avoidance direction and the obstacle avoidance reverse direction are opposite, and both the obstacle avoidance direction and the obstacle avoidance reverse direction are one of the multiple search directions; the step of determining the obstacle avoidance direction and the obstacle avoidance reverse direction based on the body density value and the neighboring density value specifically comprises: calculating density differences in the multiple search directions based on the body density value and the neighboring density value; the density difference is the difference between the body density value and the neighboring density value in the corresponding search direction; selecting the search direction with the largest density difference as the obstacle avoidance direction, and determining the search direction opposite to the obstacle avoidance direction as the obstacle avoidance reverse direction; Obtaining an obstacle avoidance rehearsal space of an adjacent UAV in the direction of the obstacle avoidance; the step of obtaining the obstacle avoidance rehearsal space of an adjacent UAV in the direction of the obstacle avoidance comprises: sending an obstacle avoidance rehearsal request to an adjacent UAV in the direction of the obstacle avoidance; receiving obstacle avoidance cooperation data returned by the adjacent UAV in the direction of the obstacle avoidance; and calculating the obstacle avoidance rehearsal space based on the obstacle position and the obstacle avoidance cooperation data; Determine an obstacle avoidance adjustment path based on the obstacle avoidance rehearsal space, the obstacle position, the obstacle avoidance direction, and the preset flight mission, and calculate the flight path of the main UAV; A flight control instruction for the main UAV is generated based on the flight path to control the main UAV to perform the preset flight mission.

2. The method according to claim 1, characterized in that The step of calculating the obstacle avoidance rehearsal space based on the obstacle position and the obstacle avoidance coordination data specifically includes: Determine the current flight state and expected flight trajectory of the adjacent UAV in the obstacle avoidance direction based on the obstacle avoidance cooperation data; Calculating an available obstacle avoidance space for the main UAV in a direction of travel based on the current flight state, the predicted flight trajectory, and the obstacle position; An obstacle avoidance rehearsal space is determined based on the available obstacle avoidance space and the maneuverability constraints of the main UAV.

3. The method according to claim 1, characterized in that Before the steps of receiving adjacent position data in multiple search directions, determining the number of adjacent drones, and calculating the density of the drone in the search direction, the method further includes: Acquire environmental parameters in the multiple search directions; the environmental parameters include wind speed, wind direction, air pressure and temperature; Assigning weights to the multiple search directions based on the environmental parameters to obtain search direction weight coefficients; The effective detection distances in the plurality of search directions are adjusted according to the search direction weight coefficients, and the preset distance range is determined based on the effective detection distances.

4. The method according to claim 1, wherein After determining the obstacle avoidance adjustment path based on the obstacle avoidance rehearsal space, the obstacle position, the obstacle avoidance direction, and the preset flight mission, and calculating the flight path of the main UAV, the method further includes: detecting dynamic obstacles on the flight path and calculating a preset trajectory of the dynamic obstacles; Based on the preset trajectory, the obstacle avoidance adjustment path and the flight path are updated.

5. The method according to claim 4, characterized in that The step of updating the obstacle avoidance adjustment path and the flight path based on the preset trajectory specifically includes: Constructing a dynamic obstacle avoidance area based on the preset trajectory of the dynamic obstacle; Calculating a minimum safe distance between the main UAV and the dynamic obstacle avoidance area, and determining a dynamic obstacle avoidance priority based on the minimum safe distance; Adjusting the obstacle avoidance strategy according to the dynamic obstacle avoidance priority and the real-time status of the obstacle avoidance rehearsal space; The obstacle avoidance adjustment path is updated based on the obstacle avoidance strategy, and the flight path is replanned in combination with the preset flight mission.

6. A control device, characterized in that: The control device includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to cause the control device to execute the method according to any one of claims 1 to 5.

7. A computer-readable storage medium comprising instructions, characterized in that: When the instruction is executed on a control device, the control device is caused to execute the method according to any one of claims 1 to 5.

8. A computer program product, characterized in that When the computer program product is run on a control device, the control device is caused to execute the method according to any one of claims 1 to 5.

Images