Multi-machine distributed decision-making swarm route avoidance cooperative control method and system

Abstract

The invention discloses a multi-aircraft distributed decision-making swarm route avoidance cooperative control method and system, relates to the technical field of unmanned aerial vehicle formation control, and is used for multi-aircraft flight formation. Each unmanned aerial vehicle is provided with an intelligent decision-making module with environment sensing, communication and path planning capabilities; the method comprises the steps that each unmanned aerial vehicle generates a preliminary avoidance path by adopting an improved genetic algorithm based on acquired local environment information and received state information of neighborhood unmanned aerial vehicles in combination with a preset avoidance rule and a preset multi-objective optimization function; the optimization objectives of the preset multi-objective optimization function comprise an obstacle avoidance safety distance, an energy consumption coefficient, a formation retention degree and task timeliness; space-time conflict detection is executed based on the multiple preliminary avoidance paths, the preliminary avoidance paths are optimized based on conflict detection results, a final conflict-free optimal path is obtained and broadcasted to other unmanned aerial vehicles in the formation, and distributed collaborative decision making is achieved; according to the invention, the control efficiency of multi-unmanned aerial vehicle formation control is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle formation control, in particular to a multi-aircraft distributed decision-making swarm route avoidance cooperative control method and system. BACKGROUND

[0002] With the development of unmanned aerial vehicle swarm technology, cooperative avoidance control of formation in complex environments faces higher requirements.

[0003] Traditional methods mostly use a master node for centralized decision-making, which is responsible for collecting and processing data from each unmanned aerial vehicle to generate avoidance strategies and flight instructions.

[0004] However, with the rapid expansion of unmanned aerial vehicle formation, the bottleneck of centralized decision-making is increasingly prominent, such as single-point failure risk. Once the master node fails, the entire formation cooperation will immediately collapse. In addition, since all data needs to be transmitted back to the central node, this puts a great pressure on the communication bandwidth in a super large-scale cluster, and the decision-making delay also increases significantly, making it difficult to meet the real-time obstacle avoidance requirements, resulting in a low overall control efficiency of multi-unmanned aerial vehicle formation, and making it difficult to meet the real-time cooperative requirements in large-scale and high-dynamic scenarios. SUMMARY

[0005] In order to improve the control efficiency of multi-unmanned aerial vehicle formation control and meet the real-time cooperative requirements in large-scale and high-dynamic scenarios, the present application provides a multi-aircraft distributed decision-making swarm route avoidance cooperative control method and system.

[0006] In the first aspect, the application aims to achieve the following technical solutions: A multi-aircraft distributed decision-making swarm route avoidance cooperative control method applied to a flight formation composed of multiple unmanned aerial vehicles; each unmanned aerial vehicle is equipped with an intelligent decision-making module with environmental perception, communication and path planning capabilities; The method comprises: Each unmanned aerial vehicle exchanges its flight state information and surrounding environment information through ad hoc network communication; Each unmanned aerial vehicle generates a preliminary avoidance path based on the acquired local environment information, received state information of neighboring unmanned aerial vehicles, combined with a preset avoidance rule and a preset multi-objective optimization function; the optimization objectives of the preset multi-objective optimization function include obstacle avoidance safety distance, energy consumption coefficient, formation keeping degree and task timeliness; Based on the spatio-temporal conflict detection of multiple preliminary avoidance paths, a conflict detection result is obtained, and the preliminary avoidance path is optimized based on the conflict detection result to obtain a final conflict-free optimal path; Each UAV broadcasts the final conflict-free optimal path and new flight state information to other UAVs in the formation, realizing distributed collaborative decision-making.

[0007] By adopting the technical scheme, a swarm route avoidance collaborative control method based on a distributed architecture and high reliability is provided, which can meet real-time collaborative requirements in large-scale and high-dynamic scenarios. The flight state information includes flight information such as position, speed, heading, flight attitude, etc. Specifically, a dynamic topology network is constructed through ad hoc network communication to form a distributed architecture, so as to eliminate the single-point failure risk of traditional centralized control. When any node fails, the remaining UAVs can still maintain formation cooperation through local communication, which can adapt to the high-precision control and avoidance requirements of high-density swarm tasks. The improved genetic algorithm is adopted to generate a path, and the local environment information and the flight state information of the neighborhood are combined to realize rapid decision-making. The decision-making process of the application is completed locally on each UAV, avoiding the delay of data uploading, centralized processing and re-issuing, reducing communication and calculation delay, improving system response real-time performance, and being suitable for rapid avoidance in dynamic and complex environments. After the initial avoidance path is generated, the time-space conflict detection and optimization are performed, so that the global conflict-free nature of the collaborative path of multiple UAVs is achieved. Even if each UAV independently plans a path, potential time-space collision risks can be eliminated through conflict detection and feedback optimization. The application realizes the transformation from "local optimum" to "global feasibility", which is beneficial to avoid the problem of "local rationality and global conflict". Finally, information synchronization is realized through a broadcast mechanism to promote the state consistency and collaborative consistency among formation members, enhance the environmental adaptability and autonomous collaborative ability, and thus improve the control efficiency of multi-UAV formation control.

[0008] In a preferred example of the application, each UAV exchanges the local flight state information and surrounding environment information through ad hoc network communication, specifically including: Each UAV periodically broadcasts its flight state information to neighboring UAVs in a fixed time window based on a preset ad hoc network communication protocol; the flight state information at least includes current coordinates, speed, acceleration, heading angle, remaining power and fault status flag; The environment perception sensors carried by each UAV collect surrounding environment information within a preset range in real time, and the surrounding environment information includes a static obstacle coordinate set, a dynamic obstacle motion trajectory and an airspace restriction area boundary; Based on the time synchronization parameters of the ad hoc network communication, each UAV receives the flight state information broadcast by neighboring UAVs in the same time window, and analyzes the transmission integrity of the flight state information; when packet loss is detected, a retransmission request is triggered until a complete set of neighboring UAV state information is obtained; The intelligent decision module of each unmanned aerial vehicle integrates local environment sensing information and received neighbor unmanned aerial vehicle state information set to form a global sensing data set.

[0009] By adopting the above technical solution, the fixed time window guarantees information exchange period, improves information interaction timeliness, and reduces misjudgment caused by information delay; through the "static obstacle coordinate set" and "dynamic obstacle motion trajectory", combined with the "airspace restriction area", dual cognition of physical environment and rule environment is formed to avoid entering the no-fly zone or colliding with the building, so as to improve the accuracy of environment perception. Then through the "time synchronization parameter receiving based" and "packet loss triggered retransmission" mechanism, common problems such as signal attenuation, shielding and interference in ad hoc network are effectively solved, so as to ensure that the neighbor state information set is complete and available, and to avoid failure caused by information loss.

[0010] In a preferred example of the present application: the preset avoidance rules include static obstacle avoidance rules and dynamic obstacle avoidance rules; the optimization target of the preset multi-objective optimization function includes obstacle avoidance safety distance, energy consumption coefficient, formation keeping degree and task timeliness; each unmanned aerial vehicle generates a preliminary avoidance path based on the local environment information obtained, the state information of the neighbor unmanned aerial vehicle received, the preset avoidance rules and the preset multi-objective optimization function by using the improved genetic algorithm, specifically including: each unmanned aerial vehicle determines its current position and the task target position in the next period based on the global sensing data set, and generates an initial path candidate set according to the preset path discretization parameters, each initial candidate path is composed of a sequence of discrete waypoints, wherein the discrete waypoints need to meet the minimum safety distance constraint; the average deviation degree of the discrete waypoints of the entire initial candidate path from the preset formation reference route needs to meet the minimum deviation threshold constraint; The minimum static distance of each waypoint in the initial candidate path to all static obstacles is calculated, and if the minimum static distance of any initial candidate path is less than the preset minimum static safety distance threshold, the corresponding path is removed; The motion trajectory of the dynamic obstacle is predicted by Kalman filtering to obtain the predicted trajectory in the future specified period; the minimum dynamic distance of each waypoint in the initial candidate path from the predicted trajectory of the dynamic obstacle is calculated, and if the minimum dynamic distance of any initial candidate path is less than the preset minimum dynamic safety distance threshold, the corresponding path is removed to obtain an effective path candidate set; the fitness evaluation rule is constructed in combination with the preset multi-objective optimization function, the improved genetic algorithm and the linear weighted fitness function are used to optimize the effective path candidate set, and after a maximum iteration number or a convergence condition, a preliminary avoidance path set containing a sequence of waypoints, predicted flight time, safety margin and energy consumption prediction value is output.

[0011] By adopting the technical scheme, path discretization is convenient for calculation, high-risk paths are pre-filtered through minimum safety distance constraint and average deviation constraint to reduce invalid calculation amount. Through the static obstacle avoidance rule of static obstacle distance, the basic safety margin is ensured, and the fixed obstacle threat is avoided rigidly; through the dynamic obstacle avoidance rule of minimum dynamic distance, the moving obstacle of the unmanned aerial vehicle in the moving process is predicted to improve the dynamic adaptability to the time-varying environment, and the linear weighted fitness function can quantize the multi-objective weight to realize the Pareto optimal path selection and improve the data quality of the preliminary avoidance path set.

[0012] In a preferred example of the present application: the space-time conflict detection is performed based on the plurality of preliminary avoidance paths, the conflict detection result is obtained, the preliminary avoidance paths are optimized based on the conflict detection result, and the final conflict-free optimal path is obtained, specifically including: According to the obtained unmanned aerial vehicle flight speed range and task time sensitive information, the time conflict detection window and the space conflict detection threshold are dynamically set, the parameter weight of the time conflict detection window and the space conflict detection threshold is adjusted based on the formation task stage; the joint discretization of the waypoint sequence of each preliminary avoidance path is performed according to the timestamp and the spatial coordinates to obtain the discrete space-time point set of each unmanned aerial vehicle; For the discrete space-time point sets of any two unmanned aerial vehicles, the time overlap degree and the space overlap degree are calculated to determine the corresponding comprehensive conflict level; the comprehensive conflict level includes high conflict, medium conflict and low conflict; The conflict detection result set is output based on the comprehensive conflict level, the path optimization based on the conflict level is performed based on the conflict detection result set, and finally the conflict-free optimal path set is obtained.

[0013] By adopting the technical scheme, the sensitivity is adjusted according to different flight stages (such as gathering, cruising and penetration). For example, the space threshold is reduced and the time window is shortened in the penetration stage to improve safety, and the cruising stage can be appropriately relaxed to reduce calculation overhead, so that the dynamic balance of resources and safety is realized, and the flexibility and adaptability of conflict detection are improved. Through the discrete space-time point set, the conflict detection not only looks “whether in the same position”, but also looks “whether in the same position at the same time”, accurately identifies the time misalignment pass (without conflict) and simultaneous intersection (high conflict), avoids misjudgment, and realizes high-precision space-time joint detection. The present application also introduces “comprehensive conflict level” to quantify the conflict severity, and different strategies can be adopted for different levels in subsequent optimization, for example, high conflict is forced to be re-planned, medium conflict is fine-tuned, and low conflict is only warned, to avoid excessive adjustment and resource waste caused by “one-size-fits-all”.

[0014] In a preferred example of the present application: before the space-time conflict detection is performed based on the plurality of preliminary avoidance paths, the conflict detection result is obtained, the method further includes: acquire flight state information and surrounding environment information of each unmanned aerial vehicle, the flight state information including current position, current speed, target navigation point and dynamic constraint parameter, and the surrounding environment information including relative position of adjacent unmanned aerial vehicles, obstacle distribution information and airspace control region boundary; predict first predicted approaching speed of entering conflict airspace and second predicted approaching speed of adjacent avoidance point of the unmanned aerial vehicle based on the current speed, the target navigation point and the relative position of adjacent unmanned aerial vehicles, in combination with the obstacle distribution information and the airspace control region boundary; generate avoidance strategy priority reference table for adjusting avoidance maneuver strategy according to the dynamic constraint parameter, the first predicted approaching speed and the second predicted approaching speed; acquire attitude adjustment reference range representing attitude change threshold in avoidance maneuver process, and construct multi-vehicle cooperative avoidance decision model according to the avoidance strategy priority reference table and the attitude adjustment reference range; acquire actual approaching speed of the unmanned aerial vehicle in real time, input the actual approaching speed into the multi-vehicle cooperative avoidance decision model, and generate avoidance decision instruction including lateral deflection angle increment and longitudinal adjustment amount.

[0015] By adopting the technical solution, forward-looking avoidance decision is realized, the first predicted approaching speed (entering conflict region) and the second predicted approaching speed (adjacent avoidance point) are predicted, the conflict severity and avoidance opportunity are predicted in advance, the passive response is changed into active prediction, the avoidance success rate is improved, the maximum deflection angle, climbing / descending rate and other physical limits that can be borne by the unmanned aerial vehicle are defined through the attitude adjustment reference range, it is ensured that the lateral deflection angle increment and the longitudinal adjustment amount generated are within the scope allowed by the body dynamics, and instability caused by excessive instruction is avoided. The multi-vehicle cooperative avoidance decision model of the application comprehensively considers the state of the unmanned aerial vehicle and the information of adjacent unmanned aerial vehicles, the avoidance instruction generated considers possible actions of adjacent unmanned aerial vehicles, the "shock avoidance" phenomenon caused by continuous approaching due to "you left me left" is avoided, and the cooperative efficiency is improved.

[0016] In a preferred example of the application, the prediction of the first predicted approaching speed of entering conflict airspace and the second predicted approaching speed of adjacent avoidance point of the unmanned aerial vehicle based on the current speed, the target navigation point and the relative position of adjacent unmanned aerial vehicles, in combination with the obstacle distribution information and the airspace control region boundary, comprises: determine initial approaching speed of entering conflict airspace of the unmanned aerial vehicle without avoidance action according to a navigation included angle between the current position, the current speed and the target navigation point, and the relative position of adjacent unmanned aerial vehicles; acquire track deflectable angle adjustment coefficient according to constraint strength of the track by the obstacle distribution information and the airspace control region boundary; According to the track deflectable angle adjustment coefficient and the initial approach speed, a first predicted approach speed when the UAV enters the conflict airspace is corrected; According to the dynamic constraint parameter, the distance between the target waypoint and the adjacent avoidance point, the current speed and the first predicted approach speed, and in combination with an air disturbance compensation amount, a second predicted approach speed when the UAV reaches the adjacent avoidance point is predicted.

[0017] By adopting the technical scheme, firstly, the initial approach speed of the UAV without performing any avoidance action is calculated based on the navigation angle between the current position, the current speed and the target waypoint, in combination with the relative position of the adjacent UAV, so as to accurately reflect the natural intersection trend of the UAV with the potential conflict source under the current flight intention. The track deflectable angle adjustment coefficient is introduced in the application, which is determined by the obstacle distribution information and the airspace control region boundary, and is used to quantify the limitation degree of the flight environment on the maneuvering freedom. For example, in a high-rise dense area or an airspace limited zone, the deflectable angle is small, the adjustment coefficient is low, and it is indicated that the avoidance space is limited; and in an open airspace, the adjustment coefficient is high, the avoidance flexibility is large, and by applying the coefficient to the initial approach speed, the dynamic correction of the first predicted approach speed is realized, so that the effective approach situation under the actual environmental constraint is more truly reflected.

[0018] In a preferred example of the application, the avoidance strategy priority reference table for adjusting the avoidance maneuvering strategy is generated according to the dynamic constraint parameter, the first predicted approach speed and the second predicted approach speed, specifically comprising: According to the distance between the target waypoint and the adjacent avoidance point and the second predicted approach speed, a predicted avoidance response time is calculated; According to the preset standard avoidance response time and the predicted avoidance response time, an avoidance time margin difference value is obtained; According to the maximum yaw acceleration and the maximum climb rate in the dynamic constraint parameter, in combination with the second predicted approach speed, a theoretical minimum avoidance maneuvering time consumption is calculated; The avoidance time margin difference value is compared with the theoretical minimum avoidance maneuvering time consumption, an avoidance urgency level is generated, a speed difference value of the first predicted approach speed and the second predicted approach speed is calculated, and an approach acceleration for predicting the dynamic change trend in the avoidance stage is obtained; According to the avoidance urgency level and the approach acceleration, a mapping relationship between the avoidance strategy execution priority and the maneuvering amplitude is established; Based on the mapping relationship, an avoidance strategy priority reference table for adjusting the avoidance maneuvering strategy is generated.

[0019] By adopting the technical scheme, the accuracy of the approaching speed prediction is improved, the initial approaching speed is calculated based on the navigation angle and the relative position, and reflects the natural intersection trend without avoidance; the flight path deflectable angle adjustment coefficient is introduced to reflect the limitation of the environment on the maneuvering freedom (for example, the deflectable angle is small in a canyon), and the first predicted approaching speed after correction is more in line with the actual flight constraints. The adjustment coefficient of the application is determined by the obstacle distribution and the airspace boundary, so that the system can automatically identify high-risk areas such as narrow channels and no-fly zone enclosures, and enhance the adaptability to complex environments. The second predicted approaching speed considers the kinetic constraints, distance, current speed, and air disturbance compensation, fully considers the actual response capability of the unmanned aerial vehicle and the influence of environmental disturbance, and improves the engineering practicability of the approaching avoidance point speed prediction.

[0020] In the second aspect, the application aims to adopt the following technical scheme: A multi-aircraft distributed decision-making swarm route avoidance cooperative control system applies a multi-aircraft distributed decision-making swarm route avoidance cooperative control method as described above. The system is applied to a flight formation composed of multiple unmanned aerial vehicles. Each unmanned aerial vehicle is configured with an intelligent decision-making module with environment perception, communication and path planning capabilities. The intelligent decision-making module further includes: A state information interaction unit is configured to exchange the flight state information and surrounding environment information of the local machine through ad hoc network communication. A preliminary path generation unit is configured to generate a preliminary avoidance path based on the obtained local environment information, the received state information of the neighboring unmanned aerial vehicles, the preset avoidance rules and the preset multi-objective optimization function, and an improved genetic algorithm. The optimization objectives of the preset multi-objective optimization function include the obstacle avoidance safety distance, the energy consumption coefficient, the formation keeping degree and the task timeliness. A conflict detection and optimization unit is configured to perform spatio-temporal conflict detection on multiple preliminary avoidance paths generated by multiple unmanned aerial vehicles, obtain a conflict detection result, and perform iterative optimization on the preliminary avoidance path based on the conflict detection result to generate a final optimal path that satisfies the conflict constraint. A cooperative decision broadcasting unit is configured to broadcast the generated final optimal path and the updated flight state information to other unmanned aerial vehicle nodes in the formation through a distributed communication network, to realize closed-loop cooperative decision-making and path synchronization among multiple aircrafts.

[0021] In the third aspect, the application aims to adopt the following technical scheme: A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned multi-aircraft distributed decision-making swarm route avoidance cooperative control method are implemented.

[0022] In a fourth aspect, the application aims to achieve the technical solutions as follows: A computer readable storage medium, the computer readable storage medium stores a computer program, the computer program is executed by a processor to realize the steps of the multi-machine distributed decision-making swarm route avoidance cooperative control method.

[0023] In summary, the present application includes at least one of the following beneficial technical effects: 1. Each unmanned aerial vehicle exchanges local information through ad hoc network communication to form comprehensive environmental awareness capability, and through the form of improved genetic algorithm and multi-objective optimization function, a global suboptimal preliminary path can be generated, each unmanned aerial vehicle node can make autonomous decisions to effectively avoid single point failure, and through time and space conflict detection optimization, the risk of multi-level path intersection collision is eliminated, the conflict-free path feasibility is ensured, the avoidance capability in complex dynamic environment is enhanced, and the control efficiency of multi-unmanned aerial vehicle formation control is improved.

[0024] 2. Significantly expand the perception boundary, improve the overall situation awareness capability of the formation, and construct a cooperative perception control mode with wide range, low delay and high reliability. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 is a flowchart of a multi-machine distributed decision-making swarm route avoidance cooperative control method in an embodiment of the present application. DETAILED DESCRIPTION

[0026] The present application will be further described in detail below in conjunction with the accompanying drawings.

[0027] In an embodiment, as shown in Figure 1 The present application discloses a multi-machine distributed decision-making swarm route avoidance cooperative control method, applied to a flight formation composed of multiple unmanned aerial vehicles; each unmanned aerial vehicle is equipped with an intelligent decision-making module with environmental perception, communication and path planning capabilities; a multi-machine distributed decision-making swarm route avoidance cooperative control method, specifically comprising the following steps: S1: Each unmanned aerial vehicle exchanges flight state information and surrounding environment information through ad hoc network communication.

[0028] In this embodiment, the intelligent decision-making module includes a laser radar, a UWB positioning module, a vision sensor, an ad hoc network communication module (based on TDMA protocol) and an embedded processor. The intelligent decision-making module preinstalls a path planning algorithm library through a software program, including an improved genetic algorithm, a Kalman filter predictor and a conflict detection engine. The ad hoc network communication exchange mode is to broadcast flight state information in fixed time slots every 100ms.

[0029] Specifically, step S1 comprises: S11: Each UAV periodically broadcasts its flight state information to neighboring UAVs in a fixed time window based on a preset ad hoc network communication protocol; the flight state information at least includes current coordinates, speed, acceleration, heading angle, remaining power, and fault status flag.

[0030] In this embodiment, the ad hoc network module adopts the IEEE 802.11s protocol (i.e., an ad hoc network communication protocol), the time slot period = 100 ms, and the bandwidth = 20 MHz; the fixed time window periodic broadcast refers to that the communication time slots are divided based on the TDMA (Time Division Multiple Access) protocol, and each UAV sends a data packet in a dedicated time slot; the flight state information includes multi-dimensional parameters such as spatial motion state, energy state, and fault diagnosis flag, wherein the spatial motion state includes current coordinates, speed, acceleration, and heading angle; the energy state includes remaining power; and the fault diagnosis flag includes hardware abnormality prompts such as prompts for GPS positioning failure, motor failure, and communication failure.

[0031] Specifically, the time slot allocation formula used in the broadcast process is: Slot i = (D rone ID mod N) × T slot , wherein i is the UAV identification index; Slot i is the starting point of the dedicated communication time slot of the UAV, with the unit of ms; mod is the modulo operation (remainder distribution); D rone ID is the unique logical identification number of the UAV; N is the total number of current formation UAVs; T slot is the basic time slot length, with the default value of 10 ms; the transmission power supports dynamic adjustment, and the RSSI value feedback of neighboring UAV nodes is obtained to maintain the signal strength at -65 ± 3 dBm.

[0032] S12: The environment perception sensor carried by each UAV collects surrounding environment information in a preset range in real time, and the surrounding environment information includes a static obstacle coordinate set, a dynamic obstacle motion trajectory, and a airspace restriction area boundary.

[0033] In this embodiment, the environment perception sensor includes a laser radar for detecting static obstacles and a visual sensor for tracking dynamic targets (i.e., dynamic obstacles), wherein the visual sensor can carry a YOLOv5 algorithm to process 1080p camera data. The intelligent decision-making module is pre-installed with an airspace database of the boundary of the control area of the current flight area.

[0034] Specifically, the static obstacle coordinate set includes a set of positions of immovable objects, the dynamic obstacle motion trajectory is a position and time sequence of a moving target, and the airspace restriction area boundary is a polygon vertex coordinate of a no-fly zone. The airspace data is loaded by loading Geojson format control area data to compare the UAV coordinates with the polygon boundary in real time.

[0035] For example, the dynamic obstacle trajectory processing when dynamically tracking the target is as follows: Traj dynamic = {(t k , x k , y k )|k = 1, 2, …, n} Wherein, Traj dynamic is the motion trajectory of the dynamic obstacle, the path of a moving target (such as a bird or other unmanned aerial vehicle) is modeled by a sequence of discrete space-time points, the data type is a two-dimensional plane trajectory, and the coordinate system is a local Cartesian coordinate system (with the current position of the machine as the origin); t k is the absolute timestamp of the kth trajectory point, which is obtained by system clock synchronization (such as the PTP protocol); x k is the horizontal position (lateral) of the dynamic obstacle corresponding to the timestamp t k in the local coordinate system, which is calculated by a visual sensor: Wherein, f is the focal length, u is the pixel horizontal coordinate, c x is the x-axis image center, d is the obstacle depth, and the coordinate system is defined with the machine as the origin and the x-axis pointing in the heading direction; y k is the vertical position (vertical) of the dynamic obstacle in the local coordinate system, Wherein v is the pixel vertical coordinate, c y is the Y-axis image center; k is the trajectory point index, indicating the data sampling order, and n is the total number of points; in a dynamic target tracking algorithm (such as YOLOv5), k corresponds to the sequence number of the video frame. For example, if the camera frame rate = 30 fps, then k = 1 represents the 1st frame and k = 2 represents the 2nd frame.

[0036] S13: Based on the time synchronization parameters of the ad hoc network communication, each unmanned aerial vehicle receives the flight state information broadcast by the neighboring unmanned aerial vehicles within the same time window, and analyzes the transmission integrity of the flight state information; when packet loss is detected, a retransmission request is triggered until a complete set of neighboring unmanned aerial vehicle state information is obtained.

[0037] In this embodiment, the time synchronization parameters are based on the IEEE 1588 PTP protocol, and the clock synchronization accuracy is ≤1 μs. The transmission integrity analysis is based on CRC-32 verification and sequence number continuity detection. When there is packet loss (such as a bit error rate > 1%), retransmission is triggered based on the missing packet sequence number, the maximum number of retransmissions is 3, and after timeout, path re-planning is started.

[0038] S14: The intelligent decision-making module of each unmanned aerial vehicle integrates the local environment perception information and the received set of neighboring unmanned aerial vehicle state information to form a global perception data set.

[0039] In this embodiment, the data integration includes spatial alignment and obstacle coordinate fusion.

[0040] S2: Each unmanned aerial vehicle generates a preliminary avoidance path based on the acquired local environment information, the received state information of the neighboring unmanned aerial vehicles, in combination with a preset avoidance rule and a preset multi-objective optimization function; the optimization objectives of the preset multi-objective optimization function include an obstacle avoidance safety distance, an energy consumption coefficient, a formation keeping degree, and a task timeliness.

[0041] In this embodiment, path discretization refers to segmenting the waypoints of the path from the starting point to the target point at an interval of 5 m. The constraint conditions for generating the initial path set include that the distance between a waypoint and a static obstacle is greater than or equal to 3 m (a safety threshold), and the formation deviation is less than or equal to 15°.

[0042] Specifically, the preset avoidance rule includes a static obstacle avoidance rule and a dynamic obstacle avoidance rule, and step S2 includes: S21: Each unmanned aerial vehicle determines its current position and the position of the task target in the next period based on the global perception data set, and generates an initial path candidate set according to a preset path discretization parameter, each initial path candidate being composed of a sequence of discrete waypoints, wherein the discrete waypoints need to satisfy a minimum safety distance constraint; the average deviation of the discrete waypoints of the entire initial path candidate from a preset formation reference route needs to satisfy a minimum deviation threshold constraint.

[0043] In this embodiment, the path discretization parameter is an interval threshold (such as 5 m) for segmenting a continuous path into discrete waypoints; the minimum safety distance constraint refers to the minimum allowable distance between a waypoint and a known obstacle, and the minimum safety distance for a static obstacle is greater than or equal to 3 m, and the minimum safety distance for a dynamic obstacle is greater than or equal to 5 m. The minimum deviation threshold constraint refers to the maximum angular deviation of the path as a whole from the formation reference route, which is less than or equal to 15° in this embodiment.

[0044] Specifically, the average deviation is calculated based on the total number of waypoints by calculating the vertical distance of each waypoint from the reference route. The minimum deviation threshold constraint is less than or equal to 15 meters, and if it exceeds, the path is regenerated.

[0045] S22: Calculate the minimum static distance of each waypoint in the initial path candidate to all static obstacles, and if there is any initial path candidate whose minimum static distance is less than a preset minimum static safety distance threshold, the corresponding path is removed.

[0046] In this embodiment, the minimum static distance refers to the Euclidean distance from a waypoint to the nearest static obstacle. The safety distance threshold can be set according to the type of obstacle, for example, greater than or equal to 5 m for a building and greater than or equal to 3 m for a tree. The distance calculation can be performed based on the obstacle coordinates using a KD tree.

[0047] Specifically, the static obstacle avoidance rule constraint is that if there is any waypoint with a minimum static distance less than a preset minimum static safety distance threshold, the corresponding path is removed.

[0048] S23: predicting the motion trajectory of the dynamic obstacle by Kalman filtering to obtain a predicted trajectory in a future specified period; calculating the minimum dynamic distance between each waypoint in the initial candidate path and the predicted trajectory of the dynamic obstacle, and if the minimum dynamic distance of any initial candidate path is less than a preset minimum dynamic safety distance threshold, the corresponding path is removed to obtain an effective path candidate set.

[0049] In this embodiment, Kalman filtering prediction predicts the future position of the dynamic obstacle based on the historical trajectory. The minimum dynamic distance is the minimum space-time distance between the waypoint and the predicted trajectory of the dynamic obstacle.

[0050] Specifically, a trajectory prediction model based on Kalman filtering is established, wherein the state equation is as follows: wherein Δt is a discretization time step, and the default value is 0.1 seconds, which can be set as a specific value as needed; x k , y k are the plane coordinates of the dynamic obstacle at time k, with the unit of meters; x k+1 , y k+1 are the plane coordinates of the dynamic obstacle at time k+1; are the horizontal and vertical velocity components at time k, respectively, with the unit of m / s; is a state transition matrix, and the covariance matrix Q = diag(0.1, 0.05, 0.1, 0.05) (position noise > velocity noise); ω k is a process noise, which refers to the system uncertainty error, such as wind disturbance and target maneuver.

[0051] For example, the future 5-second trajectory (Δt = 0.1s) of a flying bird (dynamic obstacle) is predicted, assuming that the position of the flying bird is (10.2m, 5.3m), then the x component of ω k reaches 0.3m due to the sudden turning of the flying bird.

[0052] The dynamic distance D is calculated by calculating the space-time distance between the waypoint and the predicted point, and the specific calculation formula is as follows: wherein x0, y0 are the coordinates of the predicted trajectory point of the dynamic obstacle; x p , y p are the coordinates of the waypoint of the planned path of the unmanned aerial vehicle, with the unit of meters; t p is the time when the unmanned aerial vehicle is expected to arrive at the point; t0 is the predicted time when the obstacle arrives at the point; α|t p -t0| is a time difference weighting item, and α is a time weight coefficient.

[0053] S24: Construct fitness evaluation rules by combining a preset multi-objective optimization function, optimize the effective path candidate set by using an improved genetic algorithm and a linear weighted fitness function, and output a preliminary avoidance path set containing a waypoint sequence, a predicted flight time, a safety margin, and a predicted energy consumption value after a maximum iteration number or a convergence condition.

[0054] In the embodiment, the linear weighted fitness function is used to convert the multi-objective into a weighted single objective; the improved genetic algorithm introduces an adaptive mutation and an elite reservation strategy. The linear weighted fitness function is designed as: wherein d safe is the obstacle avoidance safety distance, the greater the better; E cost is the estimated energy consumption, based on the path length and acceleration change; f3 is the formation keeping degree; t mission is the task time consumption; ω1, ω2, ω3, and ω4 are weight coefficients, which can be assigned values of 0.4, 0.3, 0.2, and 0.1, respectively.

[0055] Specifically, the adaptive mutation rate formula in the genetic operator improvement of the improved genetic algorithm is: m P -0.01×gen = 0.1 x e m , wherein P filtered is the mutation probability, with a value range of 0-1 and an initial value of 0.1; 0.1 is the basic mutation probability, with a value range of 0.05-0.3; -0.01 is the decay coefficient, and gen is the iteration number of the genetic algorithm. The elite reservation strategy is to combine the roulette selection with the elite reservation strategy, and reserve the top 10% individuals in terms of fitness in each generation to directly enter the next generation. The crossover operator uses an adaptive crossover probability.

[0056] After screening through steps S21-S23, the effective path candidate set obtained is denoted as P ij , the obstacle avoidance safety distance in the preset multi-objective optimization function is d safe , the greater the value, the higher the safety; the estimated energy consumption Ecost = η∑‖v k ‖ 2 , wherein η is the power coefficient energy consumption coefficient constant; v k is the flight speed of the path segment; the formation keeping degree is f3 = ∑‖P j - P base_j ‖, wherein P mina,b is the current position coordinate of the jth UAV on the path, and P conflict is the target position coordinate of the jth UAV on the reference flight line.

[0057] S3: performing time-space conflict detection based on the plurality of preliminary avoidance paths, obtaining a conflict detection result, optimizing the preliminary avoidance paths based on the conflict detection result, and obtaining a final conflict-free optimal path.

[0058] In this embodiment, step S3 comprises: S31: dynamically setting a time conflict detection window and a space conflict detection threshold according to the obtained UAV flight speed range and task time sensitivity information, and adjusting the parameter weights of the time conflict detection window and the space conflict detection threshold based on the formation task phase.

[0059] In this embodiment, the time conflict detection window (ΔT) is used to determine the length of the time interval in which the two UAVs have a risk of overlap in the time dimension, for example, ΔT = 2s indicates that the time overlap of the two UAVs within 2s is detected, and the initial value is set to the time corresponding to the maximum braking distance of the UAV. The space conflict detection threshold is used to determine the minimum safe distance in which the two UAVs have a risk of collision in the space dimension, for example, a space conflict detection threshold of 3m indicates that when the distance between the two UAVs is less than 3m, it is determined that there is a risk of space conflict, and the initial value is set to twice the maximum size of the UAV. The task time sensitivity information refers to the degree of requirement for time accuracy in the task phase, which can be set by the task instruction or manually, for example, the “formation assembly phase” requires strict synchronization, and the “return and disperse phase” has lower requirements for time accuracy.

[0060] Specifically, the formation task phase is, for example, reconnaissance, formation, and evacuation, etc. If the task phase is “high sensitivity” (such as formation flight), the time synchronization requirement needs to be improved, ΔT is shortened to generally 70% of the initial value, and the space conflict detection threshold is reduced to 70% of the initial value, etc. The low sensitivity can be lowered.

[0061] S32: discretizing the sequence of waypoints of each preliminary avoidance path according to the timestamp and the spatial coordinates to obtain a discrete space-time point set of each UAV; Specifically, the timestamp and the corresponding spatial coordinates of each path are paired to form a discrete space-time point set.

[0062] S33: calculating the time overlap degree and the space overlap degree for the discrete space-time point sets of any two UAVs, and determining the corresponding comprehensive conflict level; the comprehensive conflict level includes high conflict, medium conflict, and low conflict.

[0063] Specifically, the time overlap degree is the overlap degree of the discrete space-time point sets S_a and S_b of the two UAVs A and B in the time dimension, and the value range is 0-1, and the value 1 means complete overlap. The space overlap degree is the overlap degree of the discrete space-time point sets of the two UAVs A and B in the space dimension, and the value range is 0-1. The comprehensive conflict level is determined according to the combination of the time overlap degree and the space overlap degree, and is divided into high conflict, medium conflict, and low conflict. The space overlap degree wherein d mina,b is the minimum spatial distance of two UAVs in the discrete spatiotemporal point set; r conflict is a spatial safety distance threshold; time overlap degree τ ab is the ratio of the length of the time intersection interval to the length of the time union interval.

[0064] high conflict is τ ab greater than 0.7 and σ ab greater than 0.5, low conflict is τ ab greater than or equal to 0.4 and less than or equal to 0.7 and σ ab greater than or equal to 0.2 and less than 0.5; otherwise, classified as a medium conflict level.

[0065] S34: outputting a conflict detection result set based on the comprehensive conflict level, performing path optimization based on the conflict level based on the conflict detection result set, and finally obtaining a conflict-free optimal path set.

[0066] In this embodiment, the conflict detection result set includes a conflict UAV pair, a conflict time, a conflict spatial region, and a conflict level information set; the path optimization strategy includes re-planning a local flight path (using an improved A* algorithm) for a high conflict path, adjusting a speed or a heading for a medium conflict path, and retaining an original path for a low conflict path according to the conflict level.

[0067] Specifically, the optimized path needs to re-perform the steps S31-S33 until the conflict level of all UAV pairs is “low conflict” or “conflict-free”, and finally output a conflict-free optimal path set.

[0068] S4: each UAV broadcasts the final conflict-free optimal path and new flight state information to other UAVs in the formation to realize distributed collaborative decision-making.

[0069] In this embodiment, the new flight state information is the updated real-time state of the UAV.

[0070] Specifically, when a UAV performs data packaging, a CRC cyclic redundancy check code is added to facilitate packet loss detection by the UAV receiving the information. In the entire formation, the UAVs regularly broadcast the packaged data to other UAVs in the formation through the wireless communication module. In this embodiment, if a new obstacle (such as a sudden bird) or a task change (such as temporarily adding a reconnaissance point) occurs during the execution process, the UAV broadcasts an emergency correction instruction (including a new obstacle avoidance path or a task target) through the ad hoc network. After receiving the correction instruction, the other UAVs repeat the step S4 to form a broadcast, feedback, and re-collaboration control closed loop.

[0071] In an embodiment, before step S3, i.e. before performing the time-space conflict detection based on the plurality of preliminary avoidance paths, the method further comprises: S301: Obtain flight state information and surrounding environment information of each unmanned aerial vehicle, the flight state information including current position, current speed, target waypoint and dynamic constraint parameters, and the surrounding environment information including relative position of adjacent unmanned aerial vehicles, obstacle distribution information and airspace control region boundary.

[0072] In the embodiment, the dynamic constraint parameters are physical limitations of the unmanned aerial vehicle flight, including maximum turning angle, maximum acceleration and maximum climb rate. The relative position of adjacent unmanned aerial vehicles is obtained by self-organizing network communication to obtain the relative position of the field unmanned aerial vehicle, and the relative coordinates are calculated. The obstacle distribution information includes a set of coordinates of static obstacles and a motion trajectory of dynamic obstacles.

[0073] S302: Based on the current speed, target waypoint and relative position of adjacent unmanned aerial vehicles, in combination with the obstacle distribution information and airspace control region boundary, predict a first predicted approach speed of the unmanned aerial vehicle entering the conflict airspace and a second predicted approach speed of the adjacent avoidance point.

[0074] In the embodiment, the conflict airspace is a spatial region in which the unmanned aerial vehicle may collide with other unmanned aerial vehicles or obstacles in future flight. The first predicted approach speed refers to the predicted speed of the unmanned aerial vehicle entering the conflict airspace without active avoidance action under the current flight state, reflecting the risk speed under natural flight. The second predicted approach speed is the predicted speed of the unmanned aerial vehicle at the adjacent avoidance point (such as the boundary point of the conflict airspace) considering the dynamic constraint, which is used to reflect the limit speed when avoiding.

[0075] Specifically, step S302 includes: S3021: Determine the initial approach speed of the unmanned aerial vehicle entering the conflict airspace without avoidance action according to the navigation angle between the current position, current speed and target waypoint, and the relative position of adjacent unmanned aerial vehicles.

[0076] In the embodiment, the navigation angle is the angle between the current speed vector and the direction of the target waypoint, reflecting the deviation of the heading. The initial approach speed refers to the estimated speed of entering the conflict airspace without obstacle avoidance action.

[0077] Specifically, the initial approach speed = current flight speed x cos θ x η neig , η neig is the position influence factor of the adjacent unmanned aerial vehicle, where θ is the navigation angle, the reference plane is the horizontal plane, and the position influence factor of the adjacent unmanned aerial vehicle where d neig is the position of the adjacent unmanned aerial vehicle, less than or equal to 50 m.

[0078] S3022: Obtain a track deflectable angle adjustment coefficient according to the constraint strength of the obstacle distribution information and the airspace control area boundary to the track.

[0079] In this embodiment, the constraint strength refers to the restriction level of the airspace control area boundary to the heading adjustment, and the value range is 0-1.0. The track deflectable angle adjustment coefficient β is used to quantify the parameter of the track adjustable range, and the value range is 0 to 1. The constraint strength is different based on different constraint types, for example, when the constraint type is the no-fly zone boundary, the constraint strength is 1.0, and the track deflectable angle adjustment coefficient takes the value of 0.1. When the constraint type is a high-rise building group, the constraint strength is 0.8, and the track deflectable angle adjustment coefficient takes the value of 0.3. When the constraint type is a low-altitude open airspace, the constraint strength is 0.2, and the track deflectable angle adjustment coefficient takes the value of 0.9.

[0080] S3023: Correct the first predicted approach speed when the ownship enters the conflict airspace according to the track deflectable angle adjustment coefficient and the initial approach speed.

[0081] In this embodiment, the first predicted approach speed is the corrected speed after considering the airspace constraint.

[0082] Specifically, the first predicted approach speed = initial approach speed × track deflectable angle adjustment coefficient + wind speed compensation amount, wherein the wind speed compensation amount is measured in real time by the airborne IMU. The track deflectable angle adjustment coefficient = 0.3 indicates that the speed is reduced to 30% of the original value due to airspace constraint.

[0083] S3024: Predict the second predicted approach speed when the ownship reaches the adjacent avoidance point according to the kinetic constraint parameters, the distance between the target waypoint and the adjacent avoidance point, the current speed and the first predicted approach speed, combined with the air disturbance compensation amount.

[0084] In this embodiment, the air disturbance compensation amount is the random speed fluctuation caused by atmospheric turbulence.

[0085] Specifically, the second predicted approach speed wherein vfirst is the first predicted approach speed; a is the maximum acceleration; d' is the distance from the target waypoint to the avoidance point; γ turb is the air disturbance compensation factor, and the value is, for example, 1.05. The air disturbance compensation factor is obtained based on the air disturbance compensation amount and calculated by the Von Karman atmospheric turbulence model.

[0086] S303: Generate an avoidance strategy priority comparison table for adjusting the avoidance maneuver strategy according to the kinetic constraint parameters, the first predicted approach speed and the second predicted approach speed.

[0087] In the embodiment, step S303 comprises: S3031: calculating a predicted avoidance response time according to the distance between the target waypoint and the adjacent avoidance point and the second predicted approaching speed.

[0088] In the embodiment, the adjacent avoidance point is a boundary point of the conflict area, and the target waypoint is a task target point of the UAV. The predicted avoidance response time is a theoretical flight time required for the UAV to reach the adjacent avoidance point from the current position at the second predicted approaching speed.

[0089] S3032: obtaining an avoidance time margin difference value according to the preset standard avoidance response time and the predicted avoidance response time.

[0090] In the embodiment, the avoidance time margin difference value is a difference value between the preset standard avoidance response time and the predicted avoidance response time, reflecting a gap between the actual available time and the task required time. The preset standard avoidance response time is set to 3 seconds, and the specific value can be customized according to the scene type.

[0091] S3033: calculating a theoretical minimum avoidance maneuver time consumption according to the maximum yaw acceleration and the maximum climb rate in the kinetic constraint parameter, in combination with the second predicted approaching speed.

[0092] In the embodiment, the theoretical minimum avoidance maneuver time consumption is the shortest time required for the UAV to perform the avoidance action at the maximum kinetic constraint, reflecting the maneuvering ability under the limit state.

[0093] Specifically, assuming that the avoidance target is “to complete speed adjustment and direction deflection before reaching the adjacent avoidance point”, the speed adjustment time consumption and the direction deflection time consumption need to be calculated, and the theoretical minimum avoidance maneuver time consumption takes the maximum value of the speed adjustment time consumption and the direction deflection time consumption.

[0094] S3034: comparing the avoidance time margin difference value with the theoretical minimum avoidance maneuver time consumption to generate an avoidance urgency level.

[0095] In the embodiment, the avoidance urgency level reflects the matching relationship between the time gap and the limit maneuvering ability, including high, medium and low conflict risks. When the time gap is much smaller than the limit maneuvering time, the risk is extremely high, corresponding to high urgency; when the time gap is greater than or equal to the limit maneuvering time and smaller than the emergency maneuvering time, the time gap is close to the limit, the risk is relatively high, corresponding to medium urgency; wherein the emergency maneuvering time is twice the limit maneuvering time. When the time gap is greater than or equal to the emergency maneuvering time. The time is sufficient, the risk is low, corresponding to low urgency.

[0096] S3035: calculating a speed difference value of the first predicted approaching speed and the second predicted approaching speed to obtain an approaching acceleration for predicting the dynamic change trend in the avoidance stage.

[0097] In the embodiment, the approach acceleration is the rate of change of the speed of the UAV when approaching the conflict airspace, reflecting the speed of the speed increase or attenuation. The approach acceleration is the ratio of the speed difference between the first predicted approach speed and the second predicted approach speed and the predicted avoidance response time. If the approach acceleration is negative and the absolute value is large, it means that the UAV is rapidly decelerating, which may be due to the need for emergency avoidance near the conflict area. If the approach acceleration is positive, it means that the UAV is accelerating to approach, and the collision risk with the obstacle needs to be extracted.

[0098] S3036: According to the avoidance urgency level and the approach acceleration, a mapping relationship between the avoidance strategy execution priority and the maneuver amplitude is established.

[0099] In the embodiment, the avoidance strategy execution priority is the execution order of the avoidance action, such as priority deceleration and priority turning. The maneuver amplitude is the specific parameter of the avoidance action, including the turning angle and the speed adjustment amount.

[0100] For example, the defined mapping relationship is as follows: S3037: Based on the mapping relationship, an avoidance strategy priority reference table for adjusting the avoidance maneuver strategy is generated.

[0101] Specifically, the avoidance strategy priority reference table is generated based on the mapping relationship of step S3036, and the priority reference table of the avoidance strategy of the UAV is generated, and the table content is as follows: S304: An attitude adjustment reference range representing the attitude change threshold in the avoidance maneuver process is obtained, and a multi-UAV cooperative avoidance decision model is constructed according to the avoidance strategy priority reference table and the attitude adjustment reference range.

[0102] In the embodiment, the attitude adjustment reference range is the maximum attitude change range allowed by the UAV during avoidance, including the maximum lateral deflection angle and the maximum vertical pitch angle. The multi-UAV cooperative avoidance decision model outputs specific avoidance actions based on the current state, the predicted speed and the avoidance strategy priority, and is based on a model predictive control (MPC) algorithm of reinforcement learning. The objective function is to minimize the deviation from the target waypoint and avoid collision. The constraint condition is constructed based on the attitude adjustment reference range and the dynamics constraint. The model is trained through historical conflict scene data (such as 300 simulated turning avoidance cases), and the control parameters are optimized.

[0103] S305: The actual approach speed of the UAV is obtained in real time, and the actual approach speed is input into the multi-UAV cooperative avoidance decision model to generate avoidance decision instructions containing the lateral deflection angle increment and the longitudinal adjustment amount.

[0104] In the embodiment, the actual approaching speed is the real-time flight speed of the UAV, and the real-time flight speed is acquired to solve the problem to avoid the influence of air flow on the avoidance decision.

[0105] It should be understood that the sequence numbers of the steps in the above embodiments do not mean the order of execution, and the execution order of the processes should be determined according to the functions and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the application.

[0106] In an embodiment, a multi-aircraft distributed decision-making swarm route avoidance cooperative control system is provided, which corresponds to the multi-aircraft distributed decision-making swarm route avoidance cooperative control method in the above embodiment.

[0107] A multi-aircraft distributed decision-making swarm route avoidance cooperative control system, the system is applied to a flight formation composed of multiple unmanned aerial vehicles, each unmanned aerial vehicle is configured with an intelligent decision-making module having environment perception, communication and path planning capabilities; the intelligent decision-making module further comprises: A state information interaction unit for exchanging flight state information and surrounding environment information of the local machine through ad hoc network communication; A preliminary path generation unit for generating a preliminary avoidance path based on the acquired local environment information, the received state information of the neighboring unmanned aerial vehicles, the preset avoidance rules and the preset multi-objective optimization function; the optimization objectives of the preset multi-objective optimization function include obstacle avoidance safety distance, energy consumption coefficient, formation keeping degree and task timeliness; A conflict detection and optimization unit for detecting time and space conflicts of multiple preliminary avoidance paths generated by multiple unmanned aerial vehicles, obtaining a conflict detection result, and iteratively optimizing the preliminary avoidance path based on the conflict detection result to generate a final optimal path meeting the conflict-free constraint; A cooperative decision broadcasting unit for broadcasting the generated final optimal path and the updated flight state information to other unmanned aerial vehicle nodes in the formation through a distributed communication network to realize closed-loop cooperative decision-making and path synchronization among multiple aircrafts.

[0108] The specific limitations of the multi-aircraft distributed decision-making swarm route avoidance cooperative control system can be referred to the limitations of the multi-aircraft distributed decision-making swarm route avoidance cooperative control method in the above, which will not be repeated here; each module in the above multi-aircraft distributed decision-making swarm route avoidance cooperative control system can be realized by software, hardware and their combinations; each module can be embedded in or independent of the processor in the computer device in hardware form, or stored in the memory in the computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0109] In one embodiment, a computer device is provided, which can be a server. The computer device comprises a processor, a memory, a network interface and a database connected by a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The database of the computer device is configured to store flight state information and surrounding environment information. The network interface of the computer device is configured to communicate with an external terminal through a network connection. The computer program, when executed by the processor, implements a multi-machine distributed decision-making swarm route avoidance cooperative control method.

[0110] In one embodiment, a computer device is provided, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor. The processor, when executing the computer program, implements the steps of the multi-machine distributed decision-making swarm route avoidance cooperative control method as described above.

[0111] In one embodiment, a computer readable storage medium is provided, which stores a computer program. The computer program, when executed by a processor, implements the steps of the multi-machine distributed decision-making swarm route avoidance cooperative control method as described above.

[0112] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. Any reference to memory, storage, database or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM) and memory bus dynamic RAM (RDRAM).

[0113] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of each functional unit and module is exemplified, and in actual application, the above-mentioned functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.

[0114] The above-described embodiments are only used to illustrate the technical solutions of the present application, but not limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand; it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A multi-machine distributed decision-making swarm route avoidance collaborative control method, characterized by: Applicable to flight formations consisting of multiple drones; each drone is equipped with an intelligent decision-making module with environmental perception, communication and path planning capabilities; The method comprises: Each drone exchanges its own flight status information and surrounding environment information through self-organizing network communication; Each UAV generates a preliminary avoidance path using an improved genetic algorithm based on the acquired local environment information and the status information received from neighboring UAVs, combined with preset avoidance rules and a preset multi-objective optimization function. The optimization objectives of the preset multi-objective optimization function include obstacle avoidance safety distance, energy consumption coefficient, formation maintenance, and mission timeliness. Performing spatiotemporal conflict detection based on the plurality of preliminary avoidance paths, obtaining conflict detection results, and optimizing the preliminary avoidance paths based on the conflict detection results to obtain a final conflict-free optimal path; Each drone broadcasts the final conflict-free optimal path and new flight status information to other drones in the formation to achieve distributed collaborative decision-making.

2. The multi-machine distributed decision-making swarm route avoidance collaborative control method according to claim 1, characterized in that: The drones communicate via a self-organizing network to exchange their own flight status information and surrounding environment information, specifically including: Each drone, based on a preset ad hoc network communication protocol, periodically broadcasts its own flight status information to neighboring drones in a fixed time window; the flight status information includes at least the current coordinates, speed, acceleration, heading angle, remaining battery power, and fault status flag; The environmental perception sensors carried by each drone collect real-time information about the surrounding environment within a preset range. The surrounding environment information includes the coordinate set of static obstacles, the motion trajectory of dynamic obstacles, and the boundaries of the airspace restriction zone; Based on the time synchronization parameters of the ad hoc network communication, each UAV receives the flight status information broadcast by neighboring UAVs within the same time window and analyzes the transmission integrity of the flight status information; when packet loss is detected, a retransmission request is triggered until a complete set of neighboring UAV status information is obtained; The intelligent decision-making module of each UAV integrates the local environmental perception information with the received neighboring UAV status information set to form a global perception data set.

3. The multi-machine distributed decision-making swarm route avoidance collaborative control method according to claim 2, characterized in that: The preset avoidance rules include static obstacle avoidance rules and dynamic obstacle avoidance rules; the optimization objectives of the preset multi-objective optimization function include obstacle avoidance safety distance, energy consumption coefficient, formation maintenance degree and mission timeliness; each UAV generates a preliminary avoidance path using an improved genetic algorithm based on the acquired local environment information and the received status information of neighboring UAVs, combined with the preset avoidance rules and the preset multi-objective optimization function, specifically including: Based on the global perception data set, each UAV determines its current position and the mission target position for the next time period, and generates an initial path candidate set according to preset path discretization parameters. Each initial candidate path consists of a sequence of discrete waypoints, where the discrete waypoints must meet a minimum safety distance constraint. The average deviation of the discrete waypoints of the entire initial candidate path from the preset formation reference route must meet a minimum deviation threshold constraint. Calculate the minimum static distance from each waypoint in the initial candidate path to all static obstacles. If the minimum static distance of any initial candidate path is less than the preset minimum static safety distance threshold, the corresponding path will be eliminated. The motion trajectory of dynamic obstacles is predicted through Kalman filtering to obtain the predicted trajectory for a specified period in the future. The minimum dynamic distance between each waypoint in the initial candidate path and the predicted trajectory of the dynamic obstacle is calculated. If the minimum dynamic distance of any initial candidate path is less than the preset minimum dynamic safety distance threshold, the corresponding path is eliminated to obtain the valid path candidate set. A fitness evaluation rule is constructed in combination with a preset multi-objective optimization function, and an improved genetic algorithm and a linear weighted fitness function are used to optimize the valid path candidate set. After the maximum number of iterations or convergence conditions, a preliminary avoidance path set containing waypoint sequences, estimated flight time, safety margins, and energy consumption prediction values ​​is output.

4. The multi-machine distributed decision-making swarm route avoidance collaborative control method according to claim 3 is characterized in that: The performing of spatiotemporal conflict detection based on the plurality of preliminary avoidance paths, obtaining conflict detection results, and optimizing the preliminary avoidance paths based on the conflict detection results to obtain a final conflict-free optimal path specifically includes: Dynamically set the time conflict detection window and space conflict detection threshold based on the acquired UAV flight speed range and mission time-sensitive information, and adjust the parameter weights of the time conflict detection window and space conflict detection threshold based on the formation mission stage; The waypoint sequence of each preliminary avoidance path is jointly discretized according to the timestamp and spatial coordinates to obtain the discrete space-time point set of each UAV; For any two UAVs’ discrete time and space point sets, calculate the temporal overlap and spatial overlap to determine the corresponding comprehensive conflict level; the comprehensive conflict level includes high conflict, medium conflict, and low conflict; A conflict detection result set is output based on the comprehensive conflict level, and a conflict level-based path optimization is performed based on the conflict detection result set, so as to finally obtain a conflict-free optimal path set.

5. The multi-machine distributed decision-making swarm route avoidance collaborative control method according to claim 4 is characterized in that: Before performing spatiotemporal conflict detection based on the plurality of preliminary avoidance paths and obtaining conflict detection results, the method further includes: Obtain flight status information and surrounding environment information for each UAV, including current position, current speed, target waypoint, and dynamic constraint parameters; and surrounding environment information including the relative position of neighboring UAVs, obstacle distribution information, and airspace control area boundaries. Based on the current speed, the relative position of the target waypoint and the neighboring UAV, and in combination with the obstacle distribution information and the airspace control area boundary, predict a first predicted closing speed for the aircraft entering the conflict airspace and a second predicted closing speed for the aircraft near the avoidance point; generating an avoidance strategy priority comparison table for adjusting an avoidance maneuver strategy based on the dynamic constraint parameter, the first predicted closing speed, and the second predicted closing speed; Obtaining an attitude adjustment reference range representing an attitude change threshold during an avoidance maneuver, and constructing a multi-aircraft coordinated avoidance decision model based on the avoidance strategy priority comparison table and the attitude adjustment reference range; The actual closing speed of the aircraft is obtained in real time, and the actual closing speed is input into the multi-aircraft coordinated avoidance decision model to generate an avoidance decision instruction including a lateral deflection angle increment and a longitudinal adjustment amount.

6. The multi-machine distributed decision-making swarm route avoidance collaborative control method according to claim 5, characterized in that: The method of predicting a first predicted closing speed for the aircraft entering the conflicting airspace and a second predicted closing speed for the aircraft near the avoidance point based on the current speed, the relative position of the target waypoint and the neighboring UAV, and the obstacle distribution information and the airspace control area boundary includes: Determining an initial closing velocity for the aircraft entering the conflicting airspace without taking any evasive action based on the current position, current velocity, and the navigation angle between the target waypoint and the relative position of the neighboring UAV; Obtaining a track deflection angle adjustment coefficient based on the obstacle distribution information and the constraint strength of the airspace control area boundary on the track; Correcting and obtaining a first predicted closing speed when the aircraft enters the conflict airspace based on the track deflectable angle adjustment coefficient and the initial closing speed; A second predicted closing speed when the aircraft reaches the adjacent avoidance point is predicted based on the dynamic constraint parameters, the distance between the target waypoint and the adjacent avoidance point, the current speed, and the first predicted closing speed, combined with an air turbulence compensation amount.

7. The multi-machine distributed decision-making swarm route avoidance collaborative control method according to claim 6, characterized in that: Generating, based on the dynamic constraint parameter, the first predicted closing speed, and the second predicted closing speed, an avoidance strategy priority comparison table for adjusting the avoidance maneuver strategy specifically includes: calculating a predicted avoidance response time based on the distance between the target waypoint and the adjacent avoidance point and the second predicted closing speed; Obtaining a difference in avoidance time margin based on a preset standard avoidance response time and the predicted avoidance response time; Calculating a theoretical minimum avoidance maneuver time based on the maximum yaw acceleration and the maximum climb rate in the dynamic constraint parameters and the second predicted closing speed; comparing the avoidance time margin difference with the theoretical minimum avoidance maneuver time to generate an avoidance urgency level; calculating a speed difference between the first predicted closing speed and the second predicted closing speed to obtain a closing acceleration for predicting a dynamic change trend in an avoidance phase; establishing a mapping relationship between avoidance strategy execution priority and maneuvering range according to the avoidance urgency level and the approach acceleration; Based on the mapping relationship, an avoidance strategy priority comparison table for comparing and adjusting the avoidance maneuver strategy is generated.

8. A swarm route avoidance coordinated control system with distributed decision-making by multiple aircraft, which is applied to execute the swarm route avoidance coordinated control method with distributed decision-making by multiple aircraft as claimed in any one of claims 1 to 7, characterized in that: The system is applied to a flight formation consisting of multiple drones, each of which is equipped with an intelligent decision-making module with environmental perception, communication and path planning capabilities; The intelligent decision-making module also includes: Status information exchange unit, used to exchange the aircraft's flight status information and surrounding environment information through ad hoc network communication; A preliminary path generation unit is configured to generate a preliminary avoidance path using an improved genetic algorithm based on the acquired local environment information and the received state information of neighboring UAVs, in combination with preset avoidance rules and a preset multi-objective optimization function; the optimization objectives of the preset multi-objective optimization function include obstacle avoidance safety distance, energy consumption coefficient, formation maintenance degree, and mission timeliness; a conflict detection and optimization unit, configured to perform spatiotemporal conflict detection on multiple preliminary avoidance paths generated by multiple UAVs, obtain conflict detection results, and iteratively optimize the preliminary avoidance paths based on the conflict detection results to generate a final optimal path that satisfies the conflict-free constraint; The collaborative decision-making broadcast unit is used to broadcast the generated final optimal path and updated flight status information to other UAV nodes in the formation through a distributed communication network, realizing closed-loop collaborative decision-making and path synchronization among multiple UAVs.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the multi-machine distributed decision-making swarm route avoidance collaborative control method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the multi-machine distributed decision-making swarm route avoidance collaborative control method as described in any one of claims 1 to 7 are implemented.

Images