A learning-based adaptive control method for safety of multi-UAV systems

Abstract

This invention discloses a learning-based safety adaptive control method for a multi-UAV system. The method includes: first, acquiring the reference trajectory and state of each UAV in an unknown interference environment to construct a prior model; second, compensating the prior model using an incremental Gaussian process to construct a feedback linearized control law, resulting in a feedback linearized adaptive controller; third, applying a Lyapunov stability-based safety barrier strategy to constrain the multi-UAV system based on the feedback linearized adaptive controller; fourth, constructing a quadratic programming model based on the results of the safety constraints and outputting the original control information; and fifth, applying the original control information to the environmental interaction of the multi-UAV system to perform adaptive control. This invention is highly efficient and has high real-time performance, and can be widely applied in the field of computer technology.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a learning-based safety adaptive control method for multi-UAV systems. Background Technology

[0002] With the development and popularization of micro-aircraft, more and more micro-aircraft will share limited airspace. Using multiple drones can efficiently perform complex tasks such as search and rescue. However, complex and uncertain airspace requires drones to have safe and adaptive control systems. Furthermore, drones also suffer from uncertainties such as system model errors and airflow disturbances, which seriously affect the control quality of drones, reduce flight performance, and even lead to collisions. Therefore, safe adaptive control in uncertain environments is crucial.

[0003] Safety is one of the key issues for unmanned aerial vehicle (UAV) systems. In many cases, violating safety constraints can lead to serious accidents. Uncertain disturbances can cause UAV systems to deviate from their current trajectory and collide. Therefore, safe control strategies are needed to ensure accurate trajectory tracking while maintaining safe flight. Furthermore, uncertain disturbances in the real world often cannot be accurately mathematically modeled, and model-based controllers perform poorly when faced with uncertain disturbances. To achieve higher control performance, controllers must be able to adapt to different uncertain disturbances.

[0004] To address the collision avoidance problem of drones, a traditional approach is to bind multiple drones to a pre-planned trajectory and centrally plan the collision-free trajectory for each drone. However, complex environmental changes often require the entire team to replan the trajectory, consuming significant computational resources and limiting the team's adaptability during task execution. It also requires the planner to update the safety trajectory frequently. Another approach is to employ distributed methods, such as the inverse buffer method, velocity barrier method, and safety barrier strategy. These methods use distributed computing to solve for the collision-free trajectory of each drone online. However, they do not consider the uncertainty of the system model and cannot guarantee the system's safety probability. Therefore, considering the model uncertainty caused by system disturbances, these methods are extended to probabilistic representations, such as using chance constraint formulas and approximate solutions with Gaussian noise. These methods consider the robot's localization and motion uncertainties. However, these methods overestimate the probability of uncertainty, leading to lower task execution efficiency.

[0005] To achieve precise control of uncertain systems, a typical approach employs inner and outer loop control. The outer loop generates the guidance law along a reference trajectory, while the inner loop performs precise motion tracking. The complex nonlinear dynamics problem can then be transformed into a set of decoupled dual integrators, using a standard linear outer-loop controller, such as a proportional-derivative controller, to achieve a decoupled linear system or feedforward linearization techniques. However, these linearization techniques rely on accurate system models, which are difficult to obtain in practice.

[0006] Another approach is to use machine learning to collect environmental disturbance data, such as employing learning-based methods to improve trajectory tracking accuracy. A Gaussian process-based model predictive control algorithm was used to solve the control problem of indeterminate dynamic equations for a quadrotor. The system requires collecting a large amount of data to build a complete system model, resulting in significant computational resources. Therefore, it is necessary to reduce the prediction range of the model predictive control. Another approach is to couple the model predictive path tracking controller with feedback control using a Gaussian process to learn the environmental disturbance model online; however, both the model predictive path tracking controller and the Gaussian process consume substantial computational resources. On a real-world airborne platform, the real-time performance of the algorithm is difficult to guarantee. Summary of the Invention

[0007] In view of this, embodiments of the present invention provide a highly efficient and real-time learning-based safety adaptive control method for multi-UAV systems.

[0008] One aspect of this invention provides a learning-based safety adaptive control method for multi-UAV systems, comprising:

[0009] Obtain the reference trajectory and state of each UAV in an unknown interference environment, and construct a prior model;

[0010] The prior model is compensated by an incremental Gaussian process to construct a feedback linearized control law, thereby obtaining a feedback linearized adaptive controller.

[0011] Based on the aforementioned feedback linearization adaptive controller, a safety barrier strategy based on Lyapunov stability is used to constrain the multi-UAV system.

[0012] Based on the results of the security constraints, a secondary program is constructed, and the original control information is output.

[0013] The original control information is applied to the environmental interaction of the multi-UAV system to perform adaptive control of the multi-UAV system.

[0014] Optionally, the step of acquiring the reference trajectory and state of each UAV in an unknown interference environment and constructing a priori model includes:

[0015] Based on the prior model and the system model error of the UAV caused by unknown environmental disturbances, the dynamics of each UAV system are described by a stochastic nonlinear discrete-time model.

[0016] Determine the reference trajectory for each UAV system, solve for the optimal control solution at the corresponding time, and obtain the optimized reference state and control variables.

[0017] Optionally, the step of compensating the prior model through an incremental Gaussian process to construct a feedback linearized control law and obtain a feedback linearized adaptive controller includes:

[0018] The current state of the drone is input into a Gaussian process model for prediction, and the predicted distribution information is obtained.

[0019] Based on the predicted distribution information, the predicted mean is calculated, and the control quantity is compensated based on the predicted mean.

[0020] Based on the optimized control quantity obtained from model predictive control, the control quantity obtained from feedback linearization, the error compensation quantity obtained from incremental Gaussian process, and the safety constraints to be designed, a feedback linearization controller is constructed.

[0021] Optionally, the step of applying a safety barrier strategy based on Lyapunov stability to the multi-UAV system according to the feedback linearization adaptive controller includes:

[0022] Based on the security domain of a multi-UAV system, define the security control space among UAV swarms;

[0023] Determine the safe control space between the drone and obstacles;

[0024] Determine the Lyapunov stability control space for controlling the stability of the unmanned aerial vehicle system.

[0025] Another aspect of this invention provides a learning-based safety adaptive control device for a multi-UAV system, comprising:

[0026] The first module is used to obtain the reference trajectory and state of each UAV in an unknown interference environment and build a prior model;

[0027] The second module is used to compensate the prior model through an incremental Gaussian process, construct a feedback linearized control law, and obtain a feedback linearized adaptive controller.

[0028] The third module is used to apply a safety barrier strategy based on Lyapunov stability to the multi-UAV system based on the feedback linearization adaptive controller.

[0029] The fourth module is used to construct a secondary program based on the results of the security constraints and output the original control information;

[0030] The fifth module is used to apply the original control information to the environmental interaction of the multi-UAV system and to perform adaptive control of the multi-UAV system.

[0031] Another aspect of the present invention provides an electronic device, including a processor and a memory;

[0032] The memory is used to store programs;

[0033] The processor executes the program to implement the method described above.

[0034] Another aspect of this invention provides a computer-readable storage medium storing a program that is executed by a processor to implement the methods described above.

[0035] This invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, causing the computer device to perform the aforementioned method.

[0036] The embodiments of this invention first acquire the reference trajectory and state of each UAV in an unknown interference environment to construct a prior model; then, the prior model is compensated using an incremental Gaussian process to construct a feedback linearized control law, resulting in a feedback linearized adaptive controller; based on the feedback linearized adaptive controller, a safety barrier strategy based on Lyapunov stability is used to apply safety constraints to the multi-UAV system; a quadratic programming algorithm is constructed based on the results of the safety constraints, and the original control information is output; finally, the original control information is applied to the environmental interaction of the multi-UAV system to perform adaptive control. This invention offers high efficiency and high real-time performance. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 The overall process flowchart provided for embodiments of the present invention. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0040] To address the problems existing in the prior art, one aspect of the present invention provides a learning-based safety adaptive control method for multi-UAV systems, comprising:

[0041] Obtain the reference trajectory and state of each UAV in an unknown interference environment, and construct a prior model;

[0042] The prior model is compensated by an incremental Gaussian process to construct a feedback linearized control law, thereby obtaining a feedback linearized adaptive controller.

[0043] Based on the aforementioned feedback linearization adaptive controller, a safety barrier strategy based on Lyapunov stability is used to constrain the multi-UAV system.

[0044] Based on the results of the security constraints, a secondary program is constructed, and the original control information is output.

[0045] The original control information is applied to the environmental interaction of the multi-UAV system to perform adaptive control of the multi-UAV system.

[0046] Optionally, the step of acquiring the reference trajectory and state of each UAV in an unknown interference environment and constructing a priori model includes:

[0047] Based on the prior model and the system model error of the UAV caused by unknown environmental disturbances, the dynamics of each UAV system are described by a stochastic nonlinear discrete-time model.

[0048] Determine the reference trajectory for each UAV system, solve for the optimal control solution at the corresponding time, and obtain the optimized reference state and control variables.

[0049] Optionally, the step of compensating the prior model through an incremental Gaussian process to construct a feedback linearized control law and obtain a feedback linearized adaptive controller includes:

[0050] The current state of the drone is input into a Gaussian process model for prediction, and the predicted distribution information is obtained.

[0051] Based on the predicted distribution information, the predicted mean is calculated, and the control quantity is compensated based on the predicted mean.

[0052] Based on the optimized control quantity obtained from model predictive control, the control quantity obtained from feedback linearization, the error compensation quantity obtained from incremental Gaussian process, and the safety constraints to be designed, a feedback linearization controller is constructed.

[0053] Optionally, the step of applying a safety barrier strategy based on Lyapunov stability to the multi-UAV system according to the feedback linearization adaptive controller includes:

[0054] Based on the security domain of a multi-UAV system, define the security control space among UAV swarms;

[0055] Determine the safe control space between the drone and obstacles;

[0056] Determine the Lyapunov stability control space for controlling the stability of the unmanned aerial vehicle system.

[0057] Another aspect of this invention provides a learning-based safety adaptive control device for a multi-UAV system, comprising:

[0058] The first module is used to obtain the reference trajectory and state of each UAV in an unknown interference environment and build a prior model;

[0059] The second module is used to compensate the prior model through an incremental Gaussian process, construct a feedback linearized control law, and obtain a feedback linearized adaptive controller.

[0060] The third module is used to apply a safety barrier strategy based on Lyapunov stability to the multi-UAV system based on the feedback linearization adaptive controller.

[0061] The fourth module is used to construct a secondary program based on the results of the security constraints and output the original control information;

[0062] The fifth module is used to apply the original control information to the environmental interaction of the multi-UAV system and to perform adaptive control of the multi-UAV system.

[0063] Another aspect of the present invention provides an electronic device, including a processor and a memory;

[0064] The memory is used to store programs;

[0065] The processor executes the program to implement the method described above.

[0066] Another aspect of this invention provides a computer-readable storage medium storing a program that is executed by a processor to implement the methods described above.

[0067] This invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, causing the computer device to perform the aforementioned method.

[0068] The specific implementation process of the present invention will now be described in detail with reference to the accompanying drawings:

[0069] Based on the limitations of existing related methods, this invention uses the predictability of model predictive control and safety barrier strategies as safety constraints to design a controller that does not require changes to the preset global trajectory. At the same time, it uses incremental Gaussian processes to improve computational efficiency, thereby achieving real-time performance of the algorithm.

[0070] This invention proposes a learning-based safety adaptive control system and method for multi-UAV systems. The method employs a unified control framework to ensure system safety and utilizes online learning to enhance the system's adaptive capability to uncertain disturbances. A safety barrier strategy based on a control Lyapunov function is used to constrain collision avoidance, ensuring provable forward invariance under uncertain conditions. Furthermore, a minimally invasive modification of the nominal control output of the trajectory tracking controller is employed to obtain a smooth control output. To achieve the above objectives, this invention includes the following steps:

[0071] Step 1: Input the reference trajectory and state of each UAV in the unknown interference environment, establish model predictive control based on the prior model, and optimize it to obtain the corrected reference state and reference control quantity.

[0072] Step 2: Compensate the prior model using an incremental Gaussian process to construct a feedback linearized control law, thereby enhancing the UAV's adaptability to uncertain environments.

[0073] Step 3: Based on the results of Step 2, apply a safety barrier strategy based on Lyapunov stability to constrain the multi-UAV system.

[0074] Step 4: Based on the results of Step 3, in order to smooth the control output, a quadratic programming is constructed based on safety constraints to select the original control output in a minimally invasive manner.

[0075] Step 5: Based on the results of Step 4, the output of the position controller is fed to the underlying attitude controller for application in the environmental interaction of the multi-UAV system.

[0076] Step Six: Repeat steps two through five until the task is completed.

[0077] The following is for reference. Figure 1Taking the safe adaptive control of a multi-UAV system in an environment with unknown interference as an example, the technical solution of the present invention will be explained in more detail.

[0078] Step 1: Consider a team consisting of N drone systems Moving within a shared 3D space. Each unmanned aerial vehicle system The dynamics are described by a stochastic nonlinear discrete-time model.

[0079]

[0080] Among them, state p i ,v i ∈R 3 The state space X is compact, and the control output... f and G are the prior models of the system, and d represents the system model error caused by unknown environmental disturbances affecting the UAV. The model we establish is based on each UAV with a value of q. i The state centered on q. i ∈R 6 and at a safe distance R i Forming a superellipsoid.

[0081] Given a reference trajectory for each unmanned aerial vehicle (UAV) system

[0082]

[0083] Where q ref (t)=[p ref (t),v ref (t)] T t0 is the start time. An MPC is constructed to solve for the problem at each sampling time t. k The open-loop finite-time optimal control problem is given by = t0 + k·dt, where... It is a discrete control cycle. At the current time t... k The specific optimal control problem to be solved is:

[0084]

[0085]

[0086] q(t)∈X

[0087]

[0088]

[0089]

[0090] To achieve precise tracking control, the optimized reference state is: The control output is u mpc

[0091] Step 2: Analyze the current state of the drone (denoted as x). * The data is input into a Gaussian process model for prediction. Consider a nonlinear model of the unmanned aerial vehicle (UAV) system. Based on the properties of the conditional distribution, the prediction distribution can be obtained as follows:

[0092]

[0093]

[0094] Using its predicted mean μ(q*), the control quantity u mpc Design of a compensation-feedback linearization controller.

[0095] For nonlinear unmanned aerial vehicle systems, let... It is a priori model with a feedback linearized control law. Designed as:

[0096]

[0097] Where u is a pseudo-control variable, which transforms the nonlinear unmanned aerial vehicle system into an approximate linear integrator model.

[0098]

[0099] Intuitively, if environmental disturbances could be accurately predicted in advance, complete compensation could be achieved, making d(q) = 0. However, usually only approximate values ​​can be obtained from limited data disturbances. To achieve precise tracking control, the pseudo-control u is designed as follows:

[0100] u = u mpc +u pd -u igp +u sbc

[0101] Among them, u mpc It is the optimal control quantity obtained from model predictive control, u pd It is the control quantity obtained by feedback linearization, u igp This is the error compensation amount obtained from the incremental Gaussian process. sbc Safety constraints to be designed

[0102] Step 3: Using the feedback linearized adaptive controller obtained in Step 2, further design safety control constraints. The safety domain S is defined by the multi-UAV system. Due to the existence of uncertain disturbances d(q) and the fact that the safety barrier strategy depends on an accurate system model, directly constructing safety constraints for the multi-UAV system using the safety barrier strategy is impossible. Based on environmental disturbances learned through incremental Gaussian processes, the safety barrier strategy is further extended to uncertain systems to ensure high-probability forward invariance of the safety domain of the multi-UAV system.

[0103] Considering collisions between drones, the safe control space among a drone swarm is defined as:

[0104]

[0105] Similarly, the safe control space between the drone and the obstacle is obtained:

[0106]

[0107] Based on this, the Lyapunov stability control space considering the stability of the unmanned aerial vehicle system is:

[0108]

[0109] Step 4: In order to smooth the control output, a quadratic programming is constructed based on safety constraints to select the original control output in a minimally invasive manner.

[0110]

[0111]

[0112] u min ≤u sbc ≤u max

[0113] in,

[0114] Step 5: Based on the results of Step 4, the output of the position controller is fed to the underlying attitude controller for application in the environmental interaction of the multi-UAV system.

[0115] Step Six: Repeat steps two through five until the task is completed.

[0116] In summary, this invention proposes a learning-based safety adaptive control method to achieve safe, accurate, and robust flight of a multi-UAV system in airspace with uncertain disturbances. First, an incremental Gaussian process is designed to learn uncertain disturbances online and provide real-time compensation for the nonlinear UAV system. Then, a safety barrier strategy with a control Lyapunov function is used to achieve scalable collision-free behavior of the multi-UAV system by modifying the nominal controller. The system's safety and high-probability asymptotic stability are demonstrated using the disturbance mean distribution provided by the incremental Gaussian process. Furthermore, model predictive control utilizes a linearized system model to optimize the trajectory, further reducing tracking errors. Ultimately, this achieves safe, accurate, and robust control of the multi-UAV system under uncertain environmental disturbances.

[0117] This invention proposes a novel control strategy for accurate trajectory tracking and safe flight of multiple unmanned aerial vehicles (UAVs) under uncertain disturbances. The proposed strategy employs unified control to ensure system safety and utilizes online learning to enhance the system's adaptive capabilities in uncertain environments.

[0118] Furthermore, this invention constrains the collision avoidance method by controlling the safety barrier strategy of the Lyapunov function to ensure provable forward invariance under uncertain conditions. Finally, a minimally invasive approach is used to modify the reference control output of the trajectory tracking controller to obtain a smooth control output.

[0119] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this invention are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is altered and sub-operations described as part of a larger operation are executed independently.

[0120] Furthermore, although the invention has been described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the described functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the invention. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional skill of an engineer. Therefore, those skilled in the art can implement the invention as set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of the invention, which is determined by the full scope of the appended claims and their equivalents.

[0121] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0122] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0123] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0124] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0125] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0126] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

[0127] The above is a detailed description of the preferred embodiments of the present invention, but the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention, and these equivalent modifications or substitutions are all included within the scope defined by the claims of this application.

Claims

1. A learning-based adaptive safety control method for multi-UAV systems, characterized in that, include: Obtain the reference trajectory and state of each UAV in an unknown interference environment, and construct a prior model; The prior model is compensated by an incremental Gaussian process to construct a feedback linearized control law, thereby obtaining a feedback linearized adaptive controller. Based on the aforementioned feedback linearization adaptive controller, a safety barrier strategy based on Lyapunov stability is used to constrain the multi-UAV system. Based on the results of the security constraints, a secondary program is constructed, and the original control information is output. The original control information is applied to the environmental interaction of the multi-UAV system to perform adaptive control of the multi-UAV system; The step of applying a safety barrier strategy based on Lyapunov stability to the multi-UAV system, according to the feedback linearization adaptive controller, includes: Based on the security domain of a multi-UAV system, define the security control space among UAV swarms; Determine the safe control space between the drone and obstacles; Determine the Lyapunov stability control space for controlling the stability of unmanned aerial vehicle systems; The safe control space among the drone swarm is defined as follows: The safe control space between the drone and the obstacle is defined as follows: The Lyapunov stability control space used to control the stability of the unmanned aerial vehicle system is defined as follows: In the formula, q represents the safe control space among the drone swarm. i Let q be the state vector of the i-th UAV. j Let j be the state vector of the j-th UAV. R is the first control input. 3 Let d be a three-dimensional real space, and d be an external perturbation. For the set of all external disturbances, Let α(·) be the first safety barrier function, and let α(·) be a positive definite function. q represents the safe control space between the drone and the obstacle. o Let be the state vector of the 0th obstacle. This is the second control input. This is the second security barrier function. This refers to the Lyapunov stability control space used to control the stability of the unmanned aerial vehicle (UAV) system. X is the third control input. r To control the feasible region of the input, V(e) is the Lyapunov function, e is the state vector error, and c is the stability constant; The quadratic programming based on the results of the security constraints is constructed using the following formula: The constraints are: in min in sbc in max In the formula, u sbc For safety constraint control, N is the total number of drones. As a reference control value, u min To control the minimum value of the input, u max To control the maximum value of the input, u mpc u is the optimal control quantity obtained from model predictive control. pd The control quantity obtained by feedback linearization, u igp This is the error compensation amount obtained from the incremental Gaussian process.

2. The safety adaptive control method for a learning-based multi-UAV system according to claim 1, characterized in that, The process of acquiring the reference trajectory and state of each UAV in an unknown interference environment and constructing a priori model includes: Based on the prior model and the system model error of the UAV caused by unknown environmental disturbances, the dynamics of each UAV system are described by a stochastic nonlinear discrete-time model. Determine the reference trajectory for each UAV system, solve for the optimal control solution at the corresponding time, and obtain the optimized reference state and control variables.

3. The safety adaptive control method for a learning-based multi-UAV system according to claim 1, characterized in that, The step of compensating the prior model through an incremental Gaussian process to construct a feedback linearized control law and obtain a feedback linearized adaptive controller includes: The current state of the drone is input into a Gaussian process model for prediction, and the predicted distribution information is obtained. Based on the predicted distribution information, the predicted mean is calculated, and the control quantity is compensated based on the predicted mean. Based on the optimized control quantity obtained from model predictive control, the control quantity obtained from feedback linearization, the error compensation quantity obtained from incremental Gaussian process, and the safety constraints to be designed, a feedback linearization controller is constructed.

4. A learning-based safety adaptive control device for a multi-UAV system, characterized in that, include: The first module is used to obtain the reference trajectory and state of each UAV in an unknown interference environment and build a prior model; The second module is used to compensate the prior model through an incremental Gaussian process, construct a feedback linearized control law, and obtain a feedback linearized adaptive controller. The third module is used to apply a safety barrier strategy based on Lyapunov stability to the multi-UAV system based on the feedback linearization adaptive controller. The fourth module is used to construct a secondary program based on the results of the security constraints and output the original control information; The fifth module is used to apply the original control information to the environmental interaction of the multi-UAV system and to perform adaptive control of the multi-UAV system. The step of applying a safety barrier strategy based on Lyapunov stability to the multi-UAV system, according to the feedback linearization adaptive controller, includes: Based on the security domain of a multi-UAV system, define the security control space among UAV swarms; Determine the safe control space between the drone and obstacles; Determine the Lyapunov stability control space for controlling the stability of unmanned aerial vehicle systems; The safe control space among the drone swarm is defined as follows: The safe control space between the drone and the obstacle is defined as follows: The Lyapunov stability control space used to control the stability of the unmanned aerial vehicle system is defined as follows: In the formula, q represents the safe control space among the drone swarm. i Let q be the state vector of the i-th UAV. j Let j be the state vector of the j-th UAV. R is the first control input. 3 Let d be a three-dimensional real space, and d be an external perturbation. For the set of all external disturbances, Let α(·) be the first safety barrier function, and let α(·) be a positive definite function. q represents the safe control space between the drone and the obstacle. o Let be the state vector of the 0th obstacle. This is the second control input. This is the second security barrier function. This refers to the Lyapunov stability control space used to control the stability of the unmanned aerial vehicle (UAV) system. X is the third control input. r To control the feasible region of the input, V(e) is the Lyapunov function, e is the state vector error, and c is the stability constant; The quadratic programming based on the results of the security constraints is constructed using the following formula: The constraints are: in min in sbc in max In the formula, u sbc For safety constraint control, N is the total number of drones. As a reference control value, u min To control the minimum value of the input, u max To control the maximum value of the input, u mpc u is the optimal control quantity obtained from model predictive control. pd The control quantity obtained by feedback linearization, u igp This is the error compensation amount obtained from the incremental Gaussian process.

5. An electronic device, characterized in that, Including the processor and memory; The memory is used to store programs; The processor executes the program to implement the method as described in any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that, The storage medium stores a program that is executed by a processor to implement the method as described in any one of claims 1 to 3.

7. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 3.

Images