A control method and system for collecting information from unmanned aerial vehicles

Abstract

The present invention provides a control method for collecting information of a drone, comprising: obtaining the initial position of the drone at the initial moment; obtaining the signal-to-noise ratio between the drone and the sensor; by defining the state variable constraints and control variable constraints of the drone, modeling the drone information collection as an optimal control problem in which both the state variables and the control variables are restricted, proposing an autonomous control method of the drone from the perspective of dynamic programming, and realizing trajectory optimization and performance improvement of the drone during information collection; by converting the state variable constraints into unrestricted state variables, and converting the control variable constraints into unrestricted control variables, it is convenient to perform iterative operations through a Q learning algorithm, and only the signal-to-noise ratio between each sensor and the drone needs to be obtained to realize strategy optimization. Based on the unrestricted state variables, unrestricted control variables and signal-to-noise ratio, the sum of the amount of information collected by the drone within a preset period is obtained, and the strategy when the sum of the amount of information obtained is the maximum value is calculated.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle communication technology, and in particular to a control method and system for collecting unmanned aerial vehicle information. Background Art

[0002] Unmanned aerial vehicles, also known as drones, can achieve full or intermittent autonomous operation through programs carried by radio remote control devices or on-board computers and other devices. Currently, drones are widely used in mobile communications, emergency communications and other fields. Based on the high maneuverability of drones, flexible deployment of communication networks can be achieved, and automatic information collection from traditional sensors can be carried out through drones.

[0003] However, it is the high maneuverability of drones that makes them face huge challenges when collecting information; when drones are running at high speeds, the distance between them and the sensor is in a constantly changing state, which will lead to a decrease in the efficiency and accuracy of signal collection.

[0004] Existing methods for optimizing UAV trajectories only simplify the movement of the UAV without considering the actual kinematic model of the UAV or the overall dynamic changes of the UAV. This causes the optimization results to deviate from the actual trajectory of the UAV and fail to perform dynamic planning based on the trajectory of the UAV. Summary of the invention

[0005] The present invention provides a control method for collecting information of an unmanned aerial vehicle, so as to solve the defects in the above-mentioned prior art.

[0006] The present invention provides a control method for collecting information of an unmanned aerial vehicle, wherein the flight area of ​​the unmanned aerial vehicle is located above a sensor cluster for collecting information, and the method is characterized by comprising:

[0007] Obtaining an initial position of the UAV at an initial time; obtaining a receiving signal-to-noise ratio between the UAV and each of the sensors;

[0008] Defining state variable constraints of the UAV, including the position coordinates of the UAV at any time, the climbing angle in the vertical direction, and the yaw angle in the horizontal direction;

[0009] Defining control variable constraints of the UAV, including the flight speed, yaw angle rate and climb angle rate of the UAV at the corresponding arbitrary time;

[0010] Converting the state variable constraints into unconstrained state variables, and converting the control variable constraints into unconstrained control variables;

[0011] Based on the unrestricted state variable, the unrestricted control variable and the signal-to-noise ratio, the sum of the information collected by the drone within a preset time period is obtained, and a strategy is calculated when the sum of the information is maximized.

[0012] A control method for collecting drone information provided by the present invention includes:

[0013] The initial position of the drone at the initial moment is S0 = (x0, y0, z0) T ;

[0014] The position of the drone at any time is S = (x, y, z) T ;

[0015] The signal-to-noise ratio between the UAV and the nth sensor is ρ n ;

[0016] Define the state variable constraint as X = (x, y, z, β, α) T ;

[0017] Define the control variable constraint as U = (v, u1, u2) T ;

[0018] The sum of the information obtained by all sensors of the drone at any time is:

[0019]

[0020] The sum of the information obtained by all sensors of the drone from the initial moment to any moment is:

[0021]

[0022] Where x0, y0, z0 are the horizontal coordinate, vertical coordinate and vertical coordinate of the UAV at the initial position respectively; x, y, z are the horizontal coordinate, vertical coordinate and vertical coordinate of the UAV at any time respectively; ρ n is the receiving signal-to-noise ratio of the UAV to the nth (n=1, 2, ..., N) sensor; β is the horizontal yaw angle of the UAV, β∈[-π, π]; α is the vertical climb angle of the UAV, α∈[-π / 2, π / 2]; u1 is the angular rate of the yaw angle; u2 is the angular rate of the climb angle; N is the total number of sensors, N>1;

[0023] Among them, h min ≤z≤h max 、a min ≤x≤a max 、b min ≤y≤b max、|u1|≤c1、|u2|≤c2;h min 、h max are the minimum and maximum vertical heights of the flight area respectively; a min 、a max are the minimum and maximum values ​​of the horizontal width of the flight area, respectively; b min 、b max are respectively the minimum and maximum values ​​of the horizontal length of the flight area; c1 is the maximum value of the angular rate of the yaw angle, and c2 is the maximum value of the angular rate of the climb angle.

[0024] Further, the state variable constraints are converted into unconstrained state variables, and the control variable constraints are converted into unconstrained control variables, including the steps of:

[0025] Get the intermediate speed variable v of the drone at any time int =v-(v max +v min ) / 2; Get the horizontal coordinate intermediate variable x int =x-(a max +a min ) / 2; Get the intermediate variable y of the ordinate int =y-(b max +b min ) / 2; Get the intermediate variable z of the vertical coordinate int = z-(h max +h min ) / 2;

[0026] Get the intermediate position variable X of the drone P，int =(x int ,y int , z int ) T ;

[0027] Further, obtain the unrestricted state variables of the drone:

[0028]

[0029]

[0030] The unrestricted position variable of the drone is obtained by converting the hyperbolic tangent function tanh x to X′ p =(x′, y′, z′) T ;

[0031] Then the unrestricted state variable of the drone is X′=(x′, y′, z′, β, α) T ;

[0032] Obtaining an intermediate control variable of the control variable includes:

[0033]

[0034] U int =(v int ,u1,u2) T ;

[0035] The unrestricted control variable is obtained by converting the hyperbolic tangent function tanh x to U′=(v′, u′1, u′2) T .

[0036] Further, obtaining the Q value function includes the following steps:

[0037] Get the unrestricted state variable of the drone at time t as X'(t) and adopt the strategy U' i (X′(t)), the sum of the information collected by the drone from time t is:

[0038]

[0039] The Q value function is:

[0040]

[0041] Among them, the strategy U′ i (X′(t)) represents the functional relationship with the unrestricted state variable X′(t) as the independent variable and the unrestricted control variable as the dependent variable; Δt is the time interval for adjusting the control variable; i≥0, representing the iterative term of the strategy; a represents the control amount adopted when the unrestricted state variable is X′(t) at time t.

[0042] Preferably, solving the optimal strategy based on the Q-value strategy iteration algorithm includes the following steps:

[0043] Set the initial strategy U′ i (X′(t)), i=0;

[0044] When the Q value function is convexly optimized for the control variable, based on the calculated Q i (X′(t), a), the iterative method to obtain the strategy is as follows:

[0045]

[0046] Let i=i+1 and execute the above iteration;

[0047] Until in, is the preset stop target, and the output U′ i+1(X′(t)) is the optimal strategy.

[0048] Preferably, the Q value is calculated by an approximate method, comprising:

[0049] The Q value function is defined as:

[0050]

[0051] in, is a polynomial function of the state variable X′ and the control variable a;

[0052] is the approximate error;

[0053] Θ i =(θ1,θ2,…,θ L ) T ;

[0054]

[0055] Perform coefficient iteration, including:

[0056]

[0057] in, is the derivative of the matrix or variable Ω with respect to time t.

[0058] When θ i After convergence, obtain Q i Approximate the value of (X′, a) to obtain the Q value;

[0059] Among them, Ψ=Φ(X′(t+Δt), U′ i )-Φ(X′(t), a); γ>0, which is the step size.

[0060] Preferably, solving the optimal strategy based on the Q-value strategy iteration algorithm includes:

[0061] When the Q-value function is non-convexly optimized with respect to the control variable, the calculation strategy U′ i+1 (X′(t)), such that for any X′(t):

[0062] Q i (X′(t),U′ i+1 )≥Q i (X′(t),U′ i );

[0063] Let i = i + 1, and execute the above iterations until At this time, the output U′ i+1 (X′(t)) is the optimal strategy.

[0064] Preferably, when the Q value function is non-convexly optimized with respect to the control variable, the strategy U′ can be updated by the gradient descent method: i+1 (X′(t)):

[0065] About to pass Q i (X′(t),U′ i+1 )≥Q i (X′(t),U′ i ) is updated to:

[0066]

[0067] Let i = i + 1, and execute the above iterations until

[0068] in, is the partial derivative of the Q-value function with respect to the unrestricted control variable; α>0 is the step size.

[0069] When the Q value is difficult to express directly, the above approximate method can be used to calculate the Q value.

[0070] The present invention also provides 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 steps of any of the above-mentioned control methods for collecting drone information are implemented.

[0071] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-mentioned control methods for collecting drone information.

[0072] The present invention provides a control method for collecting information from an unmanned aerial vehicle. By defining state variable constraints and control variable constraints of the unmanned aerial vehicle, the unmanned aerial vehicle information collection is modeled as an optimal control problem in which both the state variables and the control variables are restricted. An autonomous control method for the unmanned aerial vehicle is proposed from a dynamic programming perspective to achieve trajectory optimization and performance improvement of the unmanned aerial vehicle during information collection. By converting the state variable constraints into unrestricted state variables and converting the control variable constraints into unrestricted control variables, it is convenient to perform iterative operations through a Q learning algorithm. Only the signal-to-noise ratio between each sensor and the unmanned aerial vehicle needs to be obtained to achieve strategy optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0074] Figure 1 This is one of the flow charts of the control method for collecting drone information provided by the present invention;

[0075] Figure 2 It is a schematic diagram of the UAV provided by the present invention collecting sensor information;

[0076] Figure 3 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0077] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0078] In one embodiment of the present invention, Figure 1 As shown, the present invention provides a control method for collecting drone information, comprising the steps of:

[0079] Obtaining an initial position of the UAV at an initial time; obtaining a signal-to-noise ratio between the UAV and each of the sensors;

[0080] Defining state variable constraints of the UAV, including the position coordinates of the UAV at any time, the climbing angle in the vertical direction, and the yaw angle in the horizontal direction;

[0081] Defining control variable constraints of the UAV, including the flight speed, yaw angle rate and climb angle rate of the UAV at the corresponding arbitrary time;

[0082] Converting the state variable constraints into unconstrained state variables, and converting the control variable constraints into unconstrained control variables;

[0083] Based on the unrestricted state variable, the unrestricted control variable and the signal-to-noise ratio, the sum of the information collected by the drone within a preset time period is obtained, and a strategy is calculated when the sum of the information is maximized.

[0084] It should be noted that the controllable trajectory brought about by the high maneuverability of drones brings a new optimization dimension for information collection. The existing methods for optimizing drone trajectories only simplify the motion model of the drone, but do not consider the actual kinematic model of the drone, which leads to a large gap between the optimization results and the actual situation. The present invention starts from the kinematics and aerodynamics of drones, takes the actual drone motion equations as constraints, models drone information collection as an optimal control problem with restricted state variables and control variables, proposes an autonomous control method for drones from the perspective of dynamic programming, and realizes trajectory optimization and performance improvement of drones during information collection.

[0085] It should be noted that if Figure 2 As shown, the flight area of ​​the UAV is located above a sensor cluster to be collected. The sensor cluster includes N sensors. The sensors send the collected information to the UAV. The goal of the UAV to collect sensor information is to maximize the amount of collected information by controlling the flight trajectory of the UAV.

[0086] In one embodiment, the initial position of the drone at the initial time is obtained as S0=(x0, y0, z0) T ;

[0087] The position of the drone at any time is S = (x, y, z) T ;

[0088] The basic equation of motion of the drone is defined as:

[0089]

[0090]

[0091] Define the state variable as X = (x, y, z, β, α) T ;

[0092] Define the control variable as U = (v, u1, u2) T ;

[0093] It should be noted that the state variable and the control variable are functions related to the time t. All subsequent symbols without time variables in the present invention are universal with symbols with time variables, such as X(t) and X;

[0094] The equation of motion of the drone can be written as: The expansion is the basic equation of motion mentioned above;

[0095] Wherein, x0, y0, z0 are the horizontal coordinate, vertical coordinate and vertical coordinate of the UAV at the initial position respectively; x, y, z are the horizontal coordinate, vertical coordinate and vertical coordinate of the UAV at any time respectively; β is the horizontal yaw angle of the UAV, β∈[-π,π]; α is the vertical climb angle of the UAV, α∈[-π / 2,π / 2]; u1 is the angular rate of the yaw angle; u2 is the angular rate of the climb angle; N is the total number of sensors, N>1;

[0096] It should be noted that the flight area is limited by the terrain and buildings. The flight altitude of the drone must be higher than the height of the terrain and buildings and lower than the maximum altitude specified by aviation regulations. min ≤z≤h max The flight area is also limited in scope horizontally, subject to mission constraints or the effective range of the control link, i.e. a min ≤x≤a max 、b min ≤y≤b max ;h min 、h max are the minimum and maximum vertical heights of the flight area respectively; a min 、a max are the minimum and maximum values ​​of the horizontal width of the flight area, respectively; b min 、b max are the minimum and maximum values ​​of the horizontal length of the flight area, respectively;

[0097] It should be noted that c1 is the maximum value of the angular velocity of the yaw angle, c2 is the maximum value of the angular velocity of the climb angle; |u1|≤c1, |u2|≤c2;

[0098] Assume that the position of the nth sensor is (x n ,y n , z n ) T , then the distance between the UAV and the nth sensor is:

[0099]

[0100] Get the signal-to-noise ratio between the UAV and the nth sensor as ρ n , the signal-to-noise ratio is related to the distance d n However, the drone can estimate its received signal-to-noise ratio through the received sensor signal without knowing the sensor position;

[0101] The sum of the information obtained by all sensors of the drone at any time is:

[0102]

[0103] The sum of the information obtained by all sensors of the drone from the initial moment to any moment is:

[0104]

[0105] It should be noted that the signal-to-noise ratio between the drone and the sensor is related to the distance d n When the distance increases, the signal-to-noise ratio decreases with the increase of distance; the distance between the UAV and the sensor is obtained here only to illustrate the signal-to-noise ratio ρ n The distance d n The method of the present invention only needs to obtain the signal-to-noise ratio between the drone and the sensor, without obtaining the position of the sensor;

[0106] Based on the above steps, the control objective of the UAV is redefined as: given the initial position of the UAV and its motion equation, as well as the constraints of its state variables and control variables, the amount of sensor information received by the UAV is maximized through control, which is converted into an optimal control problem:

[0107]

[0108] st X0=(x0, y0, z0, β0, α0) T ;

[0109]

[0110]

[0111] |u1|≤c1,|u2|≤c2,v min ≤v≤v max ;

[0112] Where, Ξ=[v min , v max ]×[-π, π]×[-π / 2, π / 2] is the strategy space of the control variable. When the control variable changes, the value range of the control variable should be the above-mentioned strategy space. For this optimal control problem, the initial state value is given, and both the state variable and the control variable are constrained.

[0113] It should be noted that the above optimal control problem can be solved by the Hamilton-Jacobi-Bellman (HJB) equation, but the solution to this problem is difficult to obtain because the HJB equation is too complicated. The HJB equation is:

[0114]

[0115] in, Indicates V * The partial derivative of (X′) with respect to the state variable X′, the HJB equation will bring about a complex functional relationship when considering the motion equation of the UAV, and it is impossible to solve the optimal control strategy from the HJB equation.

[0116] It should be noted that: 1. Multiple sensors can use frequency division multiple access to send information to the drone to ensure that there is no interference between the sensors; time division multiple access can also be used. In the above expression, the amount of information from the sensor to the drone (i.e., the instantaneous channel capacity) needs to be added with 1 / N, but it does not affect the optimization solution process; 2. When the sensor sends information to the drone, it can use rateless coding (such as fountain code) and other coding methods that reach the channel capacity for transmission, and the amount of information transmitted can reach the channel capacity; 3. The integration interval in the above optimization problem is infinite. In fact, due to energy and other limitations, it is impossible for the drone to fly for an infinite amount of time. Infinite time should be understood as the drone's flight time being long enough. This understanding can also be confirmed by the following facts: the optimal flight trajectory of the drone is determined. After finding the optimal trajectory of the drone, it will continue to fly according to the optimal flight trajectory until the given time ends; 4. In the above optimization problem, it is assumed that the communication between the drone and the sensor does not change due to the drone's flight posture. In actual implementation, multiple antennas can be deployed on the drone to achieve omnidirectional reception;

[0117] Furthermore, in one embodiment, the problem of obtaining the maximum amount of information for the drone is calculated by a Q learning algorithm to obtain the best control strategy, including:

[0118] Converting the state variable constraints into unconstrained state variables and converting the control variable constraints into unconstrained control variables comprises the steps of:

[0119] Get the intermediate speed variable v of the drone at any time int =v-(v max +v min ) / 2; Get the horizontal coordinate intermediate variable x int =x-(a max +a min ) / 2; Get the intermediate variable y of the ordinate int =y-(b max +b min ) / 2; Get the intermediate variable z of the vertical coordinate int = z-(h max +h min ) / 2;

[0120] Get the intermediate position variable X of the drone P，int =(x int ,y int , z int )T ;

[0121] Further, obtain the unrestricted state variables of the drone:

[0122]

[0123] The unrestricted position variable of the drone is obtained by converting the hyperbolic tangent function tanh x to X′ p =(x′, y′, z′) T ;

[0124] Then the unrestricted state variable of the drone is X′=(x′, y′, z′, β, α) T ;

[0125] Obtaining an intermediate control variable of the control variable includes:

[0126]

[0127] U int =(v int ,u1,u2) T ;

[0128] The unrestricted control variable is obtained by converting the hyperbolic tangent function tanh x to U′=(v′, u′1, u′2) T ;

[0129] Specifically, the transformed UAV motion equation is:

[0130] in,

[0131] u1=c1 tanh u′1, u2=c2 tanh u′2;

[0132] Therefore, the transformed equation of motion is expanded as:

[0133]

[0134] It should be noted that and Both are 3×3 diagonal matrices; U′=(v′,u′1,u′2) T With U=(v,u1,u2) T There is a one-to-one correspondence, but each variable in the former is not restricted. Therefore, when optimizing the Q value, the unrestricted control variable becomes an unconstrained optimization condition, which is convenient for iterative and approximate processing in the Q learning algorithm. The new state variable X′=(x′, y′, z′, β, α)T and control variable U′=(v′,u′1,u′2) T Are not restricted.

[0135] Further, the Q value function is obtained, including:

[0136] Get the unrestricted state variable of the drone at time t as X'(t) and adopt the strategy U' i (X′(t)), the sum of the information collected by the drone from time t is:

[0137] Starting from time t, we have:

[0138]

[0139] further:

[0140]

[0141] The Q value function is:

[0142]

[0143] Its physical meaning is that when the state is X′(t) at time t, strategy a is adopted, and after reaching the state X′(t+Δt), strategy U′ is still adopted. i (X′(t+Δt)), the sum of the total information collected by the drone;

[0144] And there is Q i (X′(t+Δt),U′ i (X′(t+Δt)))=V i (X′(t+Δt));

[0145] Among them, the strategy U′ i (X′(t)) represents the functional relationship with the unrestricted state variable X′(t) as the independent variable and the unrestricted control variable as the dependent variable; Δt is the time interval for adjusting the control variable; i≥0, represents the iteration term of the strategy; a represents the control amount adopted when the unrestricted state variable is X′(t) at time t;

[0146] It should be noted that strategy U′ i (X′(t)) is related to the state variable X′(t), and the strategy U′ in the subsequent expression i The state variable X′(t) can be omitted in (X′(t)) and is expressed as U′ i ; Meanwhile, the state variable X′(t) is related to the time t, and can be omitted and represented as X′ for the convenience of expression; this is only used as a further explanation of the embodiment of the present invention, and should not be regarded as a limitation of the present invention;

[0147] In one embodiment, the optimal control problem is solved based on the Q-value policy iteration algorithm:

[0148]

[0149] st X0=(x0, y0, z0, β0, α0) T ;

[0150]

[0151] |u1|≤c1,|u2|≤c2,v min ≤v≤v max ;

[0152] include:

[0153] Set the initial strategy U′ i (X′(t)), i=0;

[0154] Solve Q for any state variable X′(t) i (X′(t), a):

[0155]

[0156] It should be noted that the autonomous control method based on the Q-learning algorithm can adopt a policy iteration method or a value iteration method;

[0157] When the Q-value function is convexly optimized for the control variable, the strategy update is transformed into:

[0158]

[0159] Let i=i+1 and execute the above iteration;

[0160] Until

[0161] in, is the preset stop target, and the output U i ′ +1 (X ′ (t)) is the optimal strategy;

[0162] It should be noted that by converting When the iterative acquisition strategy is updated, the optimal solution of the above optimal control problem can be obtained.

[0163] In one embodiment, further calculating the Q value by an approximate method and improving the strategy include:

[0164] The Q value function is defined as:

[0165]

[0166] When the number of items L is large enough, Approximately obtain Q i (X ′ ,a) value;

[0167] in, is the state variable X ′ and a polynomial function of the control variable a;

[0168] is the approximate error;

[0169] Θ i =(θ1,θ2,…,θ L ) T ;

[0170]

[0171] Perform coefficient iteration, including:

[0172]

[0173] When θ i After convergence, obtain Q i Approximate the value of (X′, a) to obtain the Q value;

[0174] Among them, Ψ=Φ(X′(t+Δt), U′ i )-Φ(X′(t), a); γ>0, is the step length;

[0175] It should be noted that represents the derivative of the matrix or variable Ω with respect to time t;

[0176] It should be noted that when the Q-value function is more complex and the expression is difficult to confirm due to the large number of variables, the strategy obtained by the above calculation method can be improved. In the solution process, it is only necessary to know the signal-to-noise ratio information received by the drone, and it is not even necessary to obtain the motion equation of the drone.

[0177] In another embodiment, when the Q-value function is non-convexly optimized with respect to the control variable, the strategy U′ is updated by gradient descent. i+1 (X′(t)):

[0178] About to pass Q i (X′(t),U′ i+1 )≥Q i (X′(t),U′ i ) is updated to:

[0179]

[0180] Let i=i+1 and execute the above iteration;

[0181] Until At this time, the output U′ i+1 (X′(t)) is the determined optimal strategy, which can obtain the local optimal solution of the optimal control problem.

[0182] in, is the partial derivative of the Q-value function with respect to the unrestricted control variable; α>0 is the step size.

[0183] When the Q value is difficult to express directly, the above approximate method can be used to calculate the Q value.

[0184] It should be noted that when the Q-value function is more complex and the expression is difficult to confirm due to the large number of variables, the above-mentioned gradient descent method can be used to improve the solved strategy, and in the solution process, only the signal-to-noise ratio information received by the drone needs to be known.

[0185] On the other hand, the present invention also provides an electronic device, such as Figure 3 As shown, the electronic device may include: a processor (processor) 310, a communication interface (Communications Interface) 320, a memory (memory) 330 and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other through the communication bus 340. The processor 310 can call the logic instructions in the memory 330 to execute a control method for collecting drone information, the method comprising: obtaining the initial position of the drone at an initial moment; obtaining the signal-to-noise ratio between the drone and each of the sensors; defining the state variable constraints of the drone, including the position coordinates of the drone at any moment, the climb angle in the vertical direction, and the yaw angle in the horizontal direction; defining the control variable constraints of the drone, including the flight speed, yaw angle angular rate, and climb angle angular rate of the drone at the corresponding any moment; converting the state variable constraints into unrestricted state variables, and converting the control variable constraints into unrestricted control variables; based on the unrestricted state variables, the unrestricted control variables, and the signal-to-noise ratio, obtaining the sum of the information collected by the drone within a preset time period, and calculating the strategy when the sum of the information is the maximum value.

[0186] In addition, the logic instructions in the above-mentioned memory 330 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0187] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute a control method for collecting unmanned aerial vehicle information, the method including: obtaining the initial position of the unmanned aerial vehicle at an initial moment; obtaining the signal-to-noise ratio between the unmanned aerial vehicle and each of the sensors; defining the state variable constraints of the unmanned aerial vehicle, including the position coordinates of the unmanned aerial vehicle at any moment, the climb angle in the vertical direction, and the yaw angle in the horizontal direction; defining the control variable constraints of the unmanned aerial vehicle, including the flight speed, yaw angle angular rate, and climb angle angular rate of the unmanned aerial vehicle at the corresponding any moment; converting the state variable constraints into unrestricted state variables, and converting the control variable constraints into unrestricted control variables; based on the unrestricted state variables, the unrestricted control variables, and the signal-to-noise ratio, obtaining the sum of the amount of information collected by the unmanned aerial vehicle within a preset time period, and calculating the strategy when the sum of the amount of information is the maximum value.

[0188] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when executed by a processor to execute a control method for collecting information from a drone, the method comprising: obtaining an initial position of the drone at an initial moment; obtaining a signal-to-noise ratio between the drone and each of the sensors; defining state variable constraints of the drone, including the position coordinates of the drone at any moment, the climb angle in the vertical direction, and the yaw angle in the horizontal direction; defining control variable constraints of the drone, including the flight speed, yaw angle angular rate, and climb angle angular rate of the drone at the corresponding any moment; converting the state variable constraints into unrestricted state variables, and converting the control variable constraints into unrestricted control variables; based on the unrestricted state variables, the unrestricted control variables, and the signal-to-noise ratio, obtaining the sum of the amounts of information collected by the drone within a preset time period, and calculating a strategy when the sum of the amounts of information is the maximum value.

[0189] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0190] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0191] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A control method for collecting information from a drone, wherein the flight area of ​​the drone is located above a sensor cluster to be collected, characterized in that: include: Get the initial position of the drone at the initial moment as S0 = (x0, y0, z0) T ; Obtain the signal-to-noise ratio ρ between the drone and each of the sensors n ; Define the state variable constraints of the drone X = (x, y, z, β, α) T , including the position coordinates S = (x, y, z) of the drone at any time T , the climb angle in the vertical direction and the yaw angle in the horizontal direction; Define the control variable constraint U of the drone = (v, u1, u2) T , including the flight speed, yaw angle rate and climb angle rate of the UAV at the corresponding any time; Converting the state variable constraints into unconstrained state variables, and converting the control variable constraints into unconstrained control variables; Based on the unrestricted state variable, the unrestricted control variable and the signal-to-noise ratio, the sum of the information collected by the drone within a preset period of time is obtained, and a strategy is calculated when the sum of the information is maximized; The sum of the instantaneous channel capacities of all sensors acquired by the UAV at any time is: The sum of the information obtained by all sensors of the drone from the initial moment to any moment is: Where x0, u0, z0 are the horizontal coordinate, vertical coordinate and vertical coordinate of the UAV at the initial position respectively; x, y, z are the horizontal coordinate, vertical coordinate and vertical coordinate of the UAV at any time respectively; ρ n is the receiving signal-to-noise ratio of the UAV to the nth sensor, which is related to the position between the UAV and the sensor; β is the horizontal yaw angle of the UAV, β∈[-π,π]; α is the vertical climb angle of the UAV, α∈[-π / 2,π / 2]; u1 is the angular rate of the yaw angle; u2 is the angular rate of the climb angle; n=1,2,…,N, N is the total number of sensors, N>1; h min ≤z≤h max 、a min ≤x≤a max 、b min ≤y≤b max 、|u1|≤c1、|u2|≤c2;h min 、h max are the minimum and maximum vertical heights of the flight area respectively; a min 、a max are the minimum and maximum values ​​of the horizontal width of the flight area, respectively; b min 、b max are respectively the minimum and maximum values ​​of the horizontal length of the flight area; c1 is the maximum value of the angular rate of the yaw angle, and c2 is the maximum value of the angular rate of the climb angle.

2. A control method for collecting drone information according to claim 1, characterized in that: Converting the state variable constraints into unconstrained state variables and converting the control variable constraints into unconstrained control variables comprises the steps of: Get the intermediate speed variable v of the drone at any time int =v-(v max +v min ) / 2; Get the horizontal coordinate intermediate variable x int =x-(a max +a min ) / 2; Get the intermediate variable y of the ordinate int =y-(b max +b min ) / 2; Get the intermediate variable z of the vertical coordinate int = z-(h max +h min ) / 2; Get the intermediate position variable X of the drone P,int =(x int ,y int ,z int ) T ; Further, obtain the unrestricted state variables of the drone: The unrestricted position variable of the drone is obtained by converting the hyperbolic tangent function tanh x to X′ p =(x′,y′,z′) T ; Then the unrestricted state variable of the drone is X′=(x′,y′,z′,β,α) T ; obtaining an intermediate control variable of the control variable, include: U int =(v int ,u1,u2) T ; The unrestricted control variable is obtained by converting the hyperbolic tangent function tanh x to U′=(v′,u′1,u′2) T .

3. The control method for collecting drone information according to claim 2, characterized in that: Get the Q value function, including: Get the unrestricted state variable of the drone at time t as X'(t) and adopt the strategy U' i (X′(t)), the sum of the information collected by the drone from time t is: The Q value function is: Among them, the strategy U′ i (X′(t)) represents the functional relationship with the unrestricted state variable as the independent variable and the unrestricted control variable as the dependent variable; Δt is the time interval for adjusting the control variable; i≥0, representing the iterative term of the strategy; a represents the control amount adopted when the unrestricted state variable is X′(t) at time t.

4. The control method for collecting drone information according to claim 3 is characterized in that: Solve the optimal strategy based on the Q-value strategy iteration algorithm, including: Set the initial strategy U′ i (X′(t)), i=0; For any state X′(t), calculate Q i (X′(t),a) is as follows: When the Q value function is convexly optimized for the control variable, based on the calculated Q i (X′(t),a), the iterative method to obtain the strategy is as follows: Let i=i+1 and execute the above iteration; Until in, is the preset stop target, and the output U′ i+1 (X′(t)) is the optimal strategy.

5. The control method for collecting drone information according to claim 4, characterized in that: The Q value is calculated by an approximate method, including: The Q value function is defined as: in, is a polynomial function of the state variable X′ and the control variable a; is the approximate error; I i =(θ1,θ2,…,θ L ) T ; Perform coefficient iteration, including: When θ i After convergence, obtain Q i Approximate the value of (X′,a) to obtain the Q value; Among them, Ψ=Φ(X′(t+Δt),U′ i )-Φ(X′(t),a);γ>0, which is the step size.

6. The control method for collecting drone information according to claim 4, characterized in that: Solve the optimal strategy based on the Q-value strategy iteration algorithm, including: When the Q-value function is non-convexly optimized with respect to the control variable, the calculation strategy U′ i+1 (X′(t)), such that for any X′(t): Q i (X′(t),U′ i+1 )≥Q i (X′(t),U′ i ); Let i = i + 1, and execute the above iteration until At this time, the output U′ i+1 (X′(t)) is the optimal strategy.

7. A method for controlling drone information collection according to claim 6, characterized in that: When the Q-value function is non-convexly optimized with respect to the control variable, the strategy U′ can be updated by the gradient descent method. i+1 (X′(t)): By Q i (X′(t),U′ i+1 )≥Q i (X′(t),U′ i ) is updated to: Let i=i+1 and execute the above iteration; Until At this time, the output U′ i+1 (X′(t)) is the optimal strategy; in, is the partial derivative of the Q-value function with respect to the unrestricted control variable; α>0 is the step size.

8. 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 steps of a control method for collecting drone information as described in any one of claims 1 to 7 are implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a control method for collecting drone information as described in any one of claims 1 to 7 are implemented.

Images