AirSim-based flight training method, device, equipment and medium

Abstract

The invention relates to a flight training method, system and device based on AirSim and a medium, and belongs to the technical field of simulation training. The method comprises the following steps: building an interference environment training platform through AirSim, presetting a cluster construction constraint condition, and obtaining a multi-machine cooperative interference array through cluster construction according to the interference environment training platform and the cluster construction constraint condition; performing anti-interference processing on the multi-machine cooperative interference array through the interference suppression dynamic model to obtain an anti-interference unmanned aerial vehicle cluster; and obtaining an anti-interference cooperative flight unmanned aerial vehicle cluster through a flight cooperative training model according to the anti-interference unmanned aerial vehicle cluster and the cluster construction constraint condition. And flight cooperative training in a failure environment is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of simulation training, and in particular relates to a flight training method, system, equipment and medium based on AirSim. Background Art

[0002] In today's rapidly developing field of drone technology, cooperative flight has become a key technology for improving drone mission capabilities and expanding their application range. However, drones often encounter various interference environments during actual flight, such as communication interruptions, sensor failures, and inclement weather. These factors pose severe challenges to cooperative flight, and traditional training methods struggle to effectively address these complex situations. Therefore, providing drone operators with efficient training in interference environments has become a key issue.

[0003] AirSim, an open-source flight simulator developed by Microsoft Research, plays a crucial role in flight training thanks to its powerful ability to simulate real-world physical environments, weather conditions, and various fault scenarios. AirSim-based flight training methods simulate various interference scenarios, allowing drones to repeatedly practice in a virtual environment, thereby improving their ability to cope with real interference conditions. Therefore, a deep integration of technological innovation and simulation training is necessary to address the urgent need for collaborative flight training for drones in interference environments and adapt to the new trends and challenges of future drone technology development. Summary of the Invention

[0004] In order to solve the above problems existing in the prior art, the present invention provides a flight training method, system, device and medium based on AirSim.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] An AirSim-based flight training method, comprising:

[0007] An interference environment training platform is built through AirSim, cluster construction constraints are preset, and a multi-machine collaborative interference array is obtained through cluster construction based on the interference environment training platform and the cluster construction constraints;

[0008] Anti-interference processing is performed on the multi-machine cooperative interference array through an interference suppression dynamic model to obtain an anti-interference UAV cluster;

[0009] An anti-interference cooperative flying UAV cluster is obtained according to the anti-interference UAV cluster and the cluster construction constraint conditions through a flight cooperative training model.

[0010] Preferably, the obtaining of a multi-machine cooperative interference array by cluster construction according to the interference environment training platform includes:

[0011] The UAV cluster is modeled by the inter-machine communication model and the cluster construction constraint conditions to obtain a UAV interaction cluster;

[0012] The inter-machine communication model includes a status information virtual broadcaster, a broadcast information virtual receiver, and a communication information virtual memory obtained through socket library programming. The status information virtual broadcaster collects drone status information, and the drone receives the drone status information through the broadcast information virtual receiver to make flight adjustments; the drone status information is decoded to obtain drone status decoding information; and the drone status decoding information is stored in the communication information virtual memory;

[0013] A multi-machine cooperative interference array is obtained by placing the UAV interactive cluster on the interference environment training platform for flight training.

[0014] Preferably, the cluster construction constraints include multi-machine dynamic state equations and multi-machine collaborative motion constraints, specifically including:

[0015] The multi-machine dynamics state equation is expressed as:

[0016]

[0017] Among them, i represents the i-th drone, represents the position vector of UAV i, represents the velocity vector of UAV i, represents the wind speed vector, represents the acceleration vector, is the acceleration correction term.

[0018] The multi-machine cooperative motion constraint condition is expressed as:

[0019]

[0020] Among them, i represents the i-th drone, j represents the j-th drone, and represents the neighboring drones of drone i. represents the velocity vector of UAV i, represents the velocity vector of UAV j, represents the position vector of UAV i, represents the position vector of UAV j, d safe represents the safe distance between UAV i and UAV j adjusted according to environmental changes, d up It represents the minimum distance between UAV i and UAV j, and f (environment factors) is the environmental factor function used to confirm the safe distance.

[0021] Preferably, performing anti-interference processing on the multi-machine cooperative interference array by using an interference suppression dynamic model to obtain an anti-interference UAV cluster includes:

[0022] Obtaining time-frequency concentration measurement values, and calculating an optimization objective function through parameter optimization according to the time-frequency concentration measurement values to obtain an optimization objective function;

[0023] Collect interference signals through the UAV sensor, and obtain a signal time-frequency diagram through time-frequency analysis based on the interference signals and the optimization objective function;

[0024] The anti-interference UAV cluster is obtained according to the signal time-frequency diagram through an adaptive interference suppression model.

[0025] Preferably, obtaining the anti-interference UAV cluster through an adaptive interference suppression model according to the signal time-frequency diagram includes:

[0026] Obtaining a signal grayscale time-frequency diagram through grayscale transformation according to the signal time-frequency diagram;

[0027] Obtaining a signal preprocessing image by threshold segmentation according to the signal grayscale time-frequency image;

[0028] Obtaining a low-frequency signal graph through a signal processing function according to the signal preprocessing graph;

[0029] The signal processing function expression is:

[0030] f p (t) = W -1 {W[y(t)]·f p (w)},

[0031] Among them, f p (t) is the signal processing function, w -1 represents inverse wavelet transform, w represents wavelet transform, y(t) is the signal preprocessing graph, f p (w) is a low-pass filter function;

[0032] Obtaining a filter function through adaptive calculation according to the low-frequency signal graph and the signal processing function;

[0033] The adaptive calculation expression is:

[0034]

[0035] Among them, f p (t) is the signal processing function, m represents the mth pulse, F p is the filtering function, t represents time, represents the rise time of the mth pulse;

[0036] An anti-interference signal is obtained by multiplying the signal preprocessing graph and the filter function, and the anti-interference UAV cluster is obtained according to the anti-interference signal.

[0037] Preferably, obtaining the anti-interference cooperative flight UAV cluster through a flight cooperative training model according to the anti-interference UAV cluster and the cluster construction constraint condition includes:

[0038] A target altitude is preset, and the anti-interference UAV cluster is controlled to fly to the target altitude by a speed control method to obtain an anti-interference UAV takeoff cluster;

[0039] Obtaining the drone's own position information and the target position information, obtaining other drone information through the communication information virtual memory, and obtaining the expected speed through speed calculation based on the drone's own position information, the target position information, and the other drone information;

[0040] A final desired speed is obtained by weighting the desired speed through a visual obstacle avoidance algorithm;

[0041] The anti-interference UAV takeoff cluster performs target search based on the target information and the final expected speed;

[0042] Obtaining the coordinates of the current target point area, storing the coordinates of the current target point area in a task information memory, and selecting the next search target point by random probability based on the coordinates of the current target point area;

[0043] A preset search tag is used to determine whether the search tag of the current target point area is 1. If yes, the drone searches for the nearest target point based on the current target point area using a quadrilateral expansion method to obtain the nearest sub-area. If no, the drone performs an area search and modifies the search tag to 1.

[0044] Determine whether the search tags are all 1. If yes, the anti-interference UAVs take off and return in a cluster; if not, repeat steps S305 to S306.

[0045] Preferably, the step of weighting the desired speed by a visual obstacle avoidance algorithm to obtain a final desired speed includes:

[0046] The depth camera collects the environment depth map, and performs gradient detection using the Sobel operator on the environment depth map to obtain an environment obstacle map;

[0047] Obtaining the target point position, and obtaining obstacle edge information through a convolutional neural network based on the environmental obstacle map;

[0048] Obtaining obstacle endpoints through endpoint detection according to the obstacle edge information;

[0049] Obtain the current position of the drone, and calculate the shortest obstacle avoidance path based on the obstacle endpoint and the current position of the drone through the shortest path calculation;

[0050] Calculating a weighted coefficient using a weighted formula according to the shortest obstacle avoidance path;

[0051] The final expected speed is obtained by comprehensive calculation according to the weighting coefficient and the expected speed.

[0052] An AirSim-based flight training system includes a training environment setup, an interference signal processing module, and a collaborative flight training module, including:

[0053] The training environment building module is used to build an interference environment training platform through AirSim, preset cluster construction constraints, and obtain a multi-machine collaborative interference array through cluster construction based on the interference environment training platform and the cluster construction constraints;

[0054] The interference signal processing module is used to perform anti-interference processing on the multi-machine cooperative interference array through an interference suppression dynamic model to obtain an anti-interference UAV cluster;

[0055] The collaborative flight training module is used to obtain an anti-interference collaborative flight UAV cluster through a flight collaborative training model according to the anti-interference UAV cluster and the cluster construction constraint conditions.

[0056] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the flight training method based on AirSim is implemented.

[0057] A storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the above-mentioned AirSim-based flight training method.

[0058] The beneficial effects of the present invention are:

[0059] (1) The UAV cluster is modeled using an inter-machine communication model and cluster construction constraints, and virtual memory is used to store and share state information. UAVs receive state information through virtual receivers and make flight adjustments, ensuring that the UAV cluster can share real-time data during collaborative flight. This improves the collaborative capabilities of the UAV swarm.

[0060] (2) By combining time-frequency analysis and adaptive filtering technology, the system can efficiently identify and suppress different types of interference. Through the adaptive interference suppression model, the suppression strategy can be dynamically adjusted according to the actual interference environment, ensuring that the UAV can fly stably in various environments. This enhances the survivability of the UAV system under environmental interference.

[0061] (3) By combining visual sensors with CNN models, the system can identify obstacle edge information in real time and make flight adjustments based on the surrounding environment to effectively avoid collisions. The final expected speed is calculated by weighting coefficients and expected speeds, and the expected speed and flight direction can be dynamically adjusted to ensure that the UAV can avoid obstacles while maintaining the efficiency and accuracy of the flight mission during flight, thereby improving the quality of flight training. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] To facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0063] Figure 1 The figure is a flow chart of a flight training method based on AirSim according to the present invention. DETAILED DESCRIPTION

[0064] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0065] See also Figure 1 , a flight training method based on AirSim, including:

[0066] S1: Using AirSim to build an interference environment training platform, presetting cluster construction constraints, and obtaining a multi-machine collaborative interference array through cluster construction based on the interference environment training platform and the cluster construction constraints;

[0067] S2: performing anti-interference processing on the multi-machine cooperative interference array through an interference suppression dynamic model to obtain an anti-interference UAV cluster;

[0068] S3: Obtain an anti-interference cooperative flying UAV cluster through a flight cooperative training model according to the anti-interference UAV cluster and the cluster construction constraint conditions.

[0069] In this embodiment, drones achieve swarming motion by exchanging position and velocity information with their neighbors. The final state of the swarm requires all drones to move in a consistent manner, and to maintain a safe distance between them while avoiding dispersal. Therefore, by defining the relative speed and position relationships between drones, the desired swarming state can be described, deriving the constraints for multi-drone coordinated motion.

[0070] In step S1, the obtaining of a multi-machine cooperative interference array by cluster construction according to the interference environment training platform includes:

[0071] S101: Modeling a drone cluster using an inter-machine communication model and the cluster construction constraints to obtain a drone interaction cluster;

[0072] S102: A multi-machine collaborative interference array is obtained by placing the UAV interactive cluster on the interference environment training platform for flight training.

[0073] Specifically, the inter-machine communication model includes:

[0074] Through socket library programming, a status information virtual broadcaster, a broadcast information virtual receiver, and a communication information virtual storage are obtained. The status information of the drone is collected according to the status information virtual broadcaster, and the drone receives the drone status information through the broadcast information virtual receiver to make flight adjustments; the drone status decoding information is obtained by decoding according to the drone status information; and the drone status decoding information is stored through the communication information virtual storage.

[0075] Specifically, the drone status information includes the drone location, mission progress, drone working status, next moment working status, next moment flight speed, and next moment location information.

[0076] In this embodiment, the virtual status information broadcaster and the virtual broadcast information receiver are programmed using the Python socket library. The address is declared as AF_INET, which specifies Internet inter-process communication. The socket type is SOCK_DGRAM, a packet socket. The socket is set to allow sending broadcast datagrams, with the level set to SOL_SOCKET, the option set to SO_BROADCAST, and the value set to 1. The virtual status information broadcaster is bound to the IP address of the computer hosting the simulation platform, and the port number is set according to the drone number. The virtual broadcast information receiver is bound to this port number. Each simulated drone has both the virtual status information broadcaster and the virtual broadcast information receiver. Before each transmission, the drone reads its latest status information from its communication information virtual memory, encrypts it into byte format, and broadcasts the datagram to the broadcast information receivers of other drones in the system. The status information broadcaster's transmission frequency is less than or equal to the receiver's receive processing frequency. The communication information virtual memory comprises three components: a receiving location memory, a location information memory, and a mission status indicator. The receiving position memory is a 3xn matrix responsible for recording the broadcasted three-dimensional position information of the drone, where n is the number of drones in the multi-drone system; the position information memory is a 3xn amplification matrix responsible for recording the three-dimensional position movement process of the drone, where n is the number of recorded coordinates; the mission status indicator changes the working status of the drone according to the completion status of the broadcast mission, and the working status includes takeoff, standby, and mission in progress.

[0077] Specifically, the cluster construction constraints include multi-machine dynamic state equations and multi-machine collaborative motion constraints;

[0078] The multi-machine dynamics state equation is expressed as:

[0079]

[0080] Among them, i represents the i-th drone, represents the position vector of UAV i, represents the velocity vector of UAV i, represents the wind speed vector, represents the acceleration vector, is the acceleration correction term.

[0081] The multi-machine cooperative motion constraint condition is expressed as:

[0082]

[0083] Among them, i represents the i-th drone, j represents the j-th drone, and represents the neighboring drones of drone i. represents the velocity vector of UAV i, represents the velocity vector of UAV j, represents the position vector of UAV i, represents the position vector of UAV j, d safe represents the safe distance between UAV i and UAV j adjusted according to environmental changes, d up It represents the minimum distance between UAV i and UAV j, and f (environment factors) is the environmental factor function used to confirm the safe distance.

[0084] In step S2, the interference suppression dynamic model includes:

[0085] S201: Obtaining time-frequency concentration measurement values, and calculating an optimization objective function through parameter optimization according to the time-frequency concentration measurement values to obtain an optimization objective function;

[0086] The parameter optimization calculation expression is:

[0087]

[0088] Among them, ρ CM is the time-frequency concentration measurement value, ξ(t) is the optimization objective function, a, b, c, d are adjustment parameters, f represents frequency, and t represents time;

[0089] S202: Collect interference signals through drone sensors, and obtain a signal time-frequency diagram through time-frequency analysis based on the interference signals and the optimization objective function. The signal time-frequency diagram includes signal amplitude changes, signal width changes, and signal frequency changes.

[0090] Specifically, the time-frequency analysis includes:

[0091]

[0092] Wherein, g(t) is the signal time-frequency diagram, x(t) is the interference signal, ξ(t) is the optimization objective function, j represents the imaginary part, and t represents time;

[0093] S203: Obtain the anti-interference UAV cluster according to the signal time-frequency diagram through an adaptive interference suppression model.

[0094] Specifically, obtaining the anti-interference UAV cluster through an adaptive interference suppression model according to the signal time-frequency diagram includes:

[0095] S203-1: Obtain a signal grayscale time-frequency diagram by grayscale transformation according to the signal time-frequency diagram;

[0096] S203-2: Obtain a signal preprocessing image by performing threshold segmentation processing on the signal grayscale time-frequency image;

[0097] S203-3: Obtain a low-frequency signal graph through a signal processing function according to the signal preprocessing graph;

[0098] The signal processing function expression is:

[0099] f p (t) = W -1 {W[y(t)]·f p (w)},

[0100] Among them, f p (t) is the signal processing function, w -1 represents inverse wavelet transform, w represents wavelet transform, y(t) is the signal preprocessing graph, f p (w) is a low-pass filter function;

[0101] S203-4: Obtaining a filter function through adaptive calculation according to the low-frequency signal graph and the signal processing function;

[0102] The adaptive calculation expression is:

[0103]

[0104] Among them, f p (t) is the signal processing function, m represents the mth pulse, F p is the filtering function, t represents time, represents the rise time of the mth pulse;

[0105] S203-5: Obtain an anti-interference signal by multiplying the signal preprocessing graph and the filter function, and obtain the anti-interference UAV cluster based on the anti-interference signal.

[0106] In step S3, the flight collaborative training model includes:

[0107] S301: Preset a target altitude, and control the anti-interference UAV cluster to fly to the target altitude by a speed control method to obtain an anti-interference UAV takeoff cluster;

[0108] S302: Obtain the drone's own position information and target position information, obtain other drone information through the communication information virtual memory, and obtain the expected speed through speed calculation based on the drone's own position information, the target position information, and the other drone information;

[0109] The speed calculation expression is:

[0110]

[0111] Among them, V goal represents the desired speed, k goal is the speed coefficient, P goal is the target location information, Ps elf The UAV's own location information;

[0112] S303: Obtaining a final desired speed by weighting the desired speed using a visual obstacle avoidance algorithm;

[0113] S304: The anti-interference UAV takeoff cluster performs target search based on the target information and the final expected speed;

[0114] S305: Acquire the coordinates of the current target point area, store the coordinates of the current target point area in the task information memory, and select the next search target point by random probability according to the coordinates of the current target point area;

[0115] S306: Preset the search tag and determine whether the search tag of the current target point area is 1. If yes, the drone searches for the nearest target point based on the current target point area using a quadrilateral expansion method to obtain the nearest sub-area. If no, the drone performs a regional search and changes the search tag to 1.

[0116] S307: Determine whether the search tags are all 1. If yes, the anti-interference UAVs take off and return in a cluster; if not, repeat steps S305 to S306.

[0117] In this embodiment, the takeoff speed is limited to 5m / s. During the search phase, the next target point is selected based on the current target position. A random probability value is used to select an unsearched area with a distance of 1 from the coordinates of the four surrounding areas as the next search target point. After the search phase is completed, the return journey begins. During this phase, the drone cluster only performs clustered movements in the vertical direction and discrete movements in the horizontal direction. A visual obstacle avoidance algorithm is also implemented to prevent collisions during the return journey. When one drone in the cluster reaches the landing trigger height and the absolute value of the height difference is less than a certain error range, it enters landing mode. Each drone returns to the airspace above the starting point and then lands. When all drones have returned, the mission is completed.

[0118] Specifically, the weighting of the desired speed by a visual obstacle avoidance algorithm to obtain a final desired speed includes:

[0119] S303-1: Collecting an environment depth map through a depth camera, and performing gradient detection using a Sobel operator on the environment depth map to obtain an environment obstacle map;

[0120] S303-2: Obtain the target point position, and obtain obstacle edge information through a convolutional neural network based on the environmental obstacle map;

[0121] S303-3: Obtaining obstacle endpoints through endpoint detection based on the obstacle edge information;

[0122] S303-4: Obtain the current position of the drone, and calculate the shortest obstacle avoidance path based on the obstacle endpoint and the current position of the drone through the shortest path calculation;

[0123] The shortest path calculation expression is:

[0124]

[0125] Among them, f final (x, y) is the path cost, x is the x-coordinate of the current position of the UAV, y is the y-coordinate of the current position of the UAV, U x is the x coordinate of the obstacle, U y is the y coordinate of the obstacle, T x is the x coordinate of the target point, T y is the y coordinate of the target point, λ is the penalty coefficient, min represents the minimum value calculation, m is the number of obstacles, d i (x, y) represents the distance from the current position of the drone to the obstacle;

[0126] S303-5: Calculating a weighted coefficient using a weighted formula according to the shortest obstacle avoidance path;

[0127] The weighted formula calculation expression is:

[0128] Gs=exp(-Ks*s),

[0129] Where Gs is the weighting coefficient used to control the balance between the target speed and the obstacle avoidance speed, Ks is the adjustment factor, s is the shortest obstacle avoidance path, and exp represents the exponential function calculation;

[0130] S303-6: Obtain the final expected speed through comprehensive calculation based on the weighting coefficient and the expected speed.

[0131] The comprehensive calculation expression is:

[0132]

[0133] Among them, V exp is the final desired speed, Gs is the weighting coefficient used to control the balance between the target speed and the obstacle avoidance speed, V goal represents the desired speed, V avoid Indicates the obstacle adjustment speed, Pself is the location information of the drone itself, i represents the i-th drone, N represents the number of drones, k avoid is the obstacle avoidance speed coefficient, P other,i Indicates the location information of all drones except drone i.

[0134] Specifically, the convolutional neural network includes 4 convolutional layers, 3 pooling layers, a fully connected layer, and an output layer; the pooling layer is located after the convolutional layer; the convolution kernel size of the first convolution layer is 3×3, and the number of convolution kernels is 32; the convolution kernel size of the second convolution layer is 3×3, and the number of convolution kernels is 64; the convolution kernel size of the third convolution layer is 3×3, and the number of convolution kernels is 128; the convolution kernel size of the fourth convolution layer is 3×3, and the number of convolution kernels is 256; the pooling layer window size is 2×2, and the stride is 2; the fully connected layer contains 256 neurons.

[0135] An AirSim-based flight training system includes a training environment setup, an interference signal processing module, and a collaborative flight training module, including:

[0136] The training environment building module is used to build an interference environment training platform through AirSim, preset cluster construction constraints, and obtain a multi-machine collaborative interference array through cluster construction based on the interference environment training platform and the cluster construction constraints;

[0137] The interference signal processing module is used to perform anti-interference processing on the multi-machine cooperative interference array through an interference suppression dynamic model to obtain an anti-interference UAV cluster;

[0138] The collaborative flight training module is used to obtain an anti-interference collaborative flight UAV cluster through a flight collaborative training model according to the anti-interference UAV cluster and the cluster construction constraint conditions.

[0139] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device.

[0140] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0141] The program code included in the computer-readable medium can be transmitted with any appropriate medium, including but not limited to wireless, electric wire, optical cable, RF or the like, or any suitable combination thereof. The computer program code for performing the operation of the present invention can be written in one or more programming languages or a combination thereof, and the programming language includes an object-oriented programming language such as Java, Smalltalk, C++, and also includes a conventional procedural programming language such as "C" language or similar programming language. The program code can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, utilizing an Internet service provider to connect through the Internet).

[0142] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as a preferred embodiment as above, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A flight training method based on AirSim, characterized in that: The following steps are involved: An interference environment training platform is built through AirSim, cluster construction constraints are preset, and a multi-machine collaborative interference array is obtained through cluster construction based on the interference environment training platform and the cluster construction constraints; Anti-interference processing is performed on the multi-machine cooperative interference array through an interference suppression dynamic model to obtain an anti-interference UAV cluster; An anti-interference cooperative flying UAV cluster is obtained according to the anti-interference UAV cluster and the cluster construction constraint conditions through a flight cooperative training model.

2. The flight training method based on AirSim according to claim 1, characterized in that: The multi-machine cooperative interference array is obtained by cluster construction according to the interference environment training platform, including: The UAV cluster is modeled by the inter-machine communication model and the cluster construction constraint conditions to obtain a UAV interaction cluster; The inter-machine communication model includes a status information virtual broadcaster, a broadcast information virtual receiver, and a communication information virtual memory obtained through socket library programming. The status information virtual broadcaster collects drone status information, and the drone receives the drone status information through the broadcast information virtual receiver to make flight adjustments; the drone status information is decoded to obtain drone status decoding information; and the drone status decoding information is stored in the communication information virtual memory; A multi-machine cooperative interference array is obtained by placing the UAV interactive cluster on the interference environment training platform for flight training.

3. The flight training method based on AirSim according to claim 1, characterized in that: The cluster construction constraints include multi-machine dynamic state equations and multi-machine collaborative motion constraints, specifically including: The multi-machine dynamics state equation is expressed as: Among them, i represents the i-th drone, represents the position vector of UAV i, represents the velocity vector of UAV i, represents the wind speed vector, represents the acceleration vector, is the acceleration correction term. The multi-machine cooperative motion constraint condition is expressed as: Among them, i represents the i-th drone, j represents the j-th drone, and represents the neighboring drones of drone i. represents the velocity vector of UAV i, represents the velocity vector of UAV j, represents the position vector of UAV i, represents the position vector of UAV j, d safe represents the safe distance between UAV i and UAV j adjusted according to environmental changes, d up It represents the minimum distance between UAV i and UAV j, and f (environment factors) is the environmental factor function used to confirm the safe distance.

4. The flight training method based on AirSim according to claim 1, characterized in that: The anti-interference processing of the multi-machine cooperative interference array by using the interference suppression dynamic model to obtain the anti-interference UAV cluster includes: Obtaining time-frequency concentration measurement values, and calculating an optimization objective function through parameter optimization according to the time-frequency concentration measurement values to obtain an optimization objective function; Collect interference signals through the UAV sensor, and obtain a signal time-frequency diagram through time-frequency analysis based on the interference signals and the optimization objective function; The anti-interference UAV cluster is obtained according to the signal time-frequency diagram through an adaptive interference suppression model.

5. The flight training method based on AirSim according to claim 4, characterized in that: The method of obtaining the anti-interference UAV cluster by using an adaptive interference suppression model according to the signal time-frequency diagram includes: Obtaining a signal grayscale time-frequency diagram through grayscale transformation according to the signal time-frequency diagram; A signal preprocessing image is obtained by threshold segmentation according to the signal grayscale time-frequency image; Obtaining a low-frequency signal graph through a signal processing function according to the signal preprocessing graph; The signal processing function expression is: f p (t)=W -1 {W[y(t)]·f p (w)}, Among them, f p (t) is the signal processing function, w -1 represents inverse wavelet transform, w represents wavelet transform, y(t) is the signal preprocessing graph, f p (w) is a low-pass filter function; Obtaining a filter function through adaptive calculation according to the low-frequency signal graph and the signal processing function; The adaptive calculation expression is: Among them, f p (t) is the signal processing function, m represents the mth pulse, F p is the filtering function, t represents time, represents the rise time of the mth pulse; An anti-interference signal is obtained by multiplying the signal preprocessing graph and the filter function, and the anti-interference UAV cluster is obtained according to the anti-interference signal.

6. The flight training method based on AirSim according to claim 1, characterized in that: The step of obtaining an anti-interference cooperative flight UAV cluster through a flight cooperative training model according to the anti-interference UAV cluster and the cluster construction constraint conditions includes: A target altitude is preset, and the anti-interference UAV cluster is controlled to fly to the target altitude by a speed control method to obtain an anti-interference UAV takeoff cluster; Obtaining the drone's own position information and the target position information, obtaining other drone information through the communication information virtual memory, and obtaining the expected speed through speed calculation based on the drone's own position information, the target position information, and the other drone information; A final desired speed is obtained by weighting the desired speed through a visual obstacle avoidance algorithm; The anti-interference UAV takeoff cluster performs target search based on the target information and the final expected speed; Obtaining the coordinates of the current target point area, storing the coordinates of the current target point area in a task information memory, and selecting the next search target point by random probability based on the coordinates of the current target point area; A preset search tag is used to determine whether the search tag of the current target point area is 1. If yes, the drone searches for the nearest target point based on the current target point area using a quadrilateral expansion method to obtain the nearest sub-area. If no, the drone performs an area search and modifies the search tag to 1. Determine whether the search tags are all 1. If yes, the anti-interference UAVs take off and return in a cluster; if not, repeat steps S305 to S306.

7. The flight training method based on AirSim according to claim 1, characterized in that: The weighting of the visual obstacle avoidance algorithm according to the expected speed to obtain the final expected speed includes: The depth camera collects the environment depth map, and the Sobel operator is used to perform gradient detection on the environment depth map to obtain the environment obstacle map; Obtaining the target point position, and obtaining obstacle edge information through a convolutional neural network based on the environmental obstacle map; Obtaining obstacle endpoints through endpoint detection according to the obstacle edge information; Obtain the current position of the drone, and calculate the shortest obstacle avoidance path based on the obstacle endpoint and the current position of the drone through the shortest path calculation; Calculating a weighted coefficient using a weighted formula according to the shortest obstacle avoidance path; The final expected speed is obtained by comprehensive calculation according to the weighting coefficient and the expected speed.

8. A flight training system based on AirSim, including a training environment setup, an interference signal processing module, and a collaborative flight training module, characterized in that: include: The training environment building module is used to build an interference environment training platform through AirSim, preset cluster construction constraints, and obtain a multi-machine collaborative interference array through cluster construction based on the interference environment training platform and the cluster construction constraints; The interference signal processing module is used to perform anti-interference processing on the multi-machine cooperative interference array through an interference suppression dynamic model to obtain an anti-interference UAV cluster; The collaborative flight training module is used to obtain an anti-interference collaborative flight UAV cluster through a flight collaborative training model according to the anti-interference UAV cluster and the cluster construction constraint conditions.

9. An electronic 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 program, the AirSim-based flight training method as described in any one of claims 1 to 7 is implemented.

10. A storage medium containing computer-executable instructions, characterized in that: When the computer executable instructions are executed by a computer processor, they are used to perform the AirSim-based flight training method according to any one of claims 1 to 7.

Images