A method for realizing autonomous drone following

Abstract

The present invention provides a method for achieving autonomous drone following, comprising: using flight time or two-way ranging to obtain ranging information between a tag at a known location and a tag at an unknown location; continuously calculating the position information of tags at unknown locations using an EKF algorithm based on the ranging information until the position information of all tags is obtained; using the point increment method to determine the optimal following position of the drone and the tags based on the position information of all tags; obtaining the coordinates of the drone's following target point based on the coordinates of the tags and the optimal following position of the drone and the tags; and obtaining the drone's real-time position using an EKF algorithm to control the drone to fly to the coordinates of the following target point. The present invention solves the problem in the prior art of immobile base station tracking positioning, which prevents autonomous dynamic following.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) positioning, and in particular to a method for realizing autonomous UAV following. Background Art

[0002] The existing conventional UWB solutions, such as TDOA (Time Difference of Arrival) and TOA (Time of Arrival), have complex system structures and cumbersome installation. The base station cannot be moved for follow-up positioning, and autonomous dynamic tracking cannot be achieved. Summary of the Invention

[0003] In order to overcome the shortcomings of the prior art, the purpose of the present invention is to provide a method for realizing autonomous drone following. The present invention solves the problem in the prior art that the base station cannot move to follow and position, and autonomous dynamic following cannot be realized.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] A method for implementing autonomous drone following, comprising:

[0006] Use time of flight or two-way ranging to obtain ranging information between a tag with a known location and a tag with an unknown location;

[0007] Based on the ranging information, the location information of the unknown location tags is continuously calculated using the EKF algorithm until the location information of all tags is obtained;

[0008] Based on the position information of all the tags, the optimal following position of the drone and the tags is solved using the point increment method;

[0009] Obtaining the coordinates of the target point followed by the drone based on the coordinates of the tag and the optimal following position of the drone and the tag;

[0010] The real-time position of the UAV is obtained through the EKF algorithm to control the UAV to fly to the coordinates of the following target point.

[0011] Preferably, obtaining the coordinates of the target point followed by the drone according to the coordinates of the tag and the optimal following position of the drone and the tag includes:

[0012] Constructing a first preset circle, where the first preset circle is a minimum circle containing the coordinate points of all tags;

[0013] Constructing a second preset circle, wherein the circumference of the second preset circle passes through the coordinate points corresponding to any three tags, or passes through only two of the three coordinate points, with the other coordinate point located inside the circle; the any three tags are the first tag, the second tag, and the third tag;

[0014] Obtaining the coordinate point of the tag corresponding to the maximum distance between the coordinate points of all tags and the center of the second preset circle to obtain the first preset coordinate point;

[0015] Determine whether the first preset coordinate point is on or within the circumference of the second preset circle. If so, obtain the corresponding equation and center radius parameter formula of the second preset circle and obtain the coordinates of the target point followed by the drone according to the corresponding equation and corresponding center radius parameter of the second preset circle. If not, construct a third preset circle until the third preset circle meets the judgment condition of the current preset circle and obtain the equation and corresponding center radius parameter corresponding to the third preset circle and obtain the coordinates of the target point followed by the drone according to the corresponding equation and corresponding center radius parameter of the third preset circle. The third preset circle is any three points of the four points including the coordinate point of the first label, the coordinate point of the second label, the coordinate point of the third label and the first preset coordinate point, and the third preset circle cannot coincide with the second preset circle.

[0016] Preferably, the second preset circle is a circumscribed circle of the coordinate points of any three tags or a circle with a diameter formed by a line connecting any two of the coordinate points of the three tags, and the other tag coordinate point is inside the circle.

[0017] Preferably, obtaining the real-time position of the UAV by using an EKF algorithm to control the UAV to fly to the target point coordinates includes:

[0018] According to the distance information between the tags, the relationship between the known coordinate points and the unknown coordinate points of the tags is constructed;

[0019] Solving the relationship using an EKF algorithm to obtain the coordinates of the unknown coordinate point;

[0020] Get the real-time location coordinates of the drone based on the coordinates of the tag;

[0021] The drone is controlled to fly to the target point coordinates according to the real-time position coordinates of the drone and the target point coordinates.

[0022] Preferably, the relationship between the known coordinate points and the unknown coordinate points of the tag is:

[0023]

[0024] Among them, x and y are the horizontal coordinate value of the unknown coordinate point and the vertical coordinate value of the unknown node respectively, d i is the distance from the i-th known node to the unknown node, x i and y i are the horizontal and vertical coordinate values ​​of the i-th known node respectively.

[0025] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0026] The present invention provides a method for realizing autonomous drone following. The present invention converts the distance relationship between the drone base station and the tag into a point set minimum circle problem, finds the optimal target point of the drone, and calculates the coordinates of the optimal target point of the drone. The real-time position of the drone is obtained through the distance information obtained by the self-organizing network device and the EKF algorithm, and the drone is controlled to fly to the coordinates of the optimal target point, thereby realizing autonomous following of the drone base station. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0028] Figure 1 A flow chart of a method for implementing autonomous drone following provided by an embodiment of the present invention;

[0029] Figure 2 An unmanned structure diagram provided for an embodiment of the present invention;

[0030] Figure 3 A schematic diagram of the positions of drones and tags provided in an embodiment of the present invention;

[0031] Figure 4 A schematic diagram of the distances between known nodes and unknown nodes provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 making creative efforts are within the scope of protection of the present invention.

[0033] The purpose of the present invention is to provide a method for realizing autonomous UAV following, which solves the problem in the prior art that the base station cannot move for following and positioning, and autonomous dynamic following cannot be realized.

[0034] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0035] like Figure 1As shown, the present invention provides a method for implementing autonomous drone following, comprising:

[0036] Step 100: Obtaining ranging information between a tag at a known location and a tag at an unknown location using time of flight or two-way ranging;

[0037] Step 200: Based on the ranging information, the location information of the unknown location tags is continuously calculated using the EKF algorithm until the location information of all tags is obtained;

[0038] Step 300: Based on the position information of all the tags, the optimal following position of the drone and the tags is solved using the point increment method;

[0039] Step 400: Obtaining the coordinates of the target point followed by the drone based on the coordinates of the tag and the optimal following position of the drone and the tag;

[0040] Step 500: Obtain the real-time position of the UAV through the EKF algorithm to control the UAV to fly to the coordinates of the tracking target point.

[0041] like Figure 2 As shown, Figure 2 This is the structure diagram of the drone. The drone consists of four parts: the main control chip MCU, the flight control unit, the 4G plus Beidou module and the BLE+UWB module.

[0042] The BLE+UWB module is responsible for measuring the distance between the base station and the tags, as well as collecting distance information between the tags. This information is then transmitted back to the main control chip, which calculates the relative positions of the tags. Combining Beidou positioning information obtained from the 4G+Beidou module, fixed-point tag location information from the cloud platform, and the flight control unit's own speed, attitude, altitude, and azimuth, it calculates the coordinates of the optimal coverage position and controls the flight control unit to fly toward the target location.

[0043] like Figure 3 As shown, based on the position information of all the tags, the point increment method is used to solve the optimal following position of the drone and the tag, including:

[0044] Obtaining a relative position vector of the drone and the tag parallel to the xy plane and a relative position vector parallel to the z axis;

[0045] Eliminate the relative position vector parallel to the z-axis.

[0046] Specifically, the tags (people) to be located are all on the ground, and their relative position vectors for distance measurement are all parallel to the xy plane. However, because the drone is in mid-air, the distance measurement information between the drone and the tag includes distance measurement information along the z-axis. Therefore, we need to decompose the distance between the person, drone, and tag into a relative position vector parallel to the xy plane and a relative position vector parallel to the z-axis. Since we only need to obtain the projection of the drone's optimal following position on the xy plane in the subsequent solution, we only retain the relative position vector parallel to the xy plane in the distance measurement equation.

[0047] Furthermore, obtaining the coordinates of the target point followed by the drone based on the coordinates of the tag and the optimal following position of the drone and the tag includes:

[0048] Constructing a first preset circle, where the first preset circle is a minimum circle containing the coordinate points of all tags;

[0049] Constructing a second preset circle, wherein the circumference of the second preset circle passes through the coordinate points corresponding to any three tags, or passes through only two of the three coordinate points, with the other coordinate point located inside the circle; the any three tags are the first tag, the second tag, and the third tag;

[0050] Obtaining the coordinate point of the tag corresponding to the maximum distance between the coordinate points of all tags and the center of the second preset circle to obtain the first preset coordinate point;

[0051] Determine whether the first preset coordinate point is on or within the circumference of the second preset circle. If so, obtain the corresponding equation and center radius parameter formula of the second preset circle and obtain the coordinates of the target point followed by the drone according to the corresponding equation and corresponding center radius parameter of the second preset circle. If not, construct a third preset circle until the third preset circle meets the judgment condition of the current preset circle and obtain the equation and corresponding center radius parameter corresponding to the third preset circle and obtain the coordinates of the target point followed by the drone according to the corresponding equation and corresponding center radius parameter of the third preset circle. The third preset circle is any three points of the four points including the coordinate point of the first label, the coordinate point of the second label, the coordinate point of the third label and the first preset coordinate point, and the third preset circle cannot coincide with the second preset circle.

[0052] The second preset circle is a circumscribed circle of the coordinate points of any three tags or a circle whose diameter is a line connecting any two of the coordinate points of the three tags, and the coordinate point of another tag is inside the circle.

[0053] Specifically, we select the best following position of the UAV. Assuming that there are N (N=1, 2, ..., Nmax) tags to be located nearby, we need to make the distance between the UAV and each tag as short as possible, so we draw a circle with the UAV as the center, hoping that this circle can contain all the tag points with the smallest radius possible. Therefore, we can simplify the problem into the minimum circle problem of a point set containing N points. The point increment method is now adopted to solve it, and its steps include the following: (1) Select any three points A, B, and C in the point set containing N points, and assume that their coordinates are (xA, yA), (xB, yB), and (xC, yC) respectively;

[0054] (2) Draw a small circle containing points A, B, and C. The circumference of the circle may pass through all three points, or it may pass through only two of them but include the third point. In the latter case, the two points on the circumference must be located at the ends of a diameter of the circle;

[0055] (3) Find the point D farthest from the center of the circle constructed in step (2) among all the points in the point set. If point D is already inside the circle or on the circumference, then the circle is the desired circle and the algorithm ends. The coordinates of the center of the circle at this time are the optimal following position of the UAV. Otherwise, execute step (4);

[0056] (4) Select three points from A, B, C, and D, and the circle generated by them is the smallest. Select three new points A1, B1, and C1 from the four points by permutation and combination (excluding the combination {A, B, C}, there are three groups in total). Return to step (2) and find the center and radius of the minimum circle corresponding to these three groups of points. Iterate until the requirements in step (3) are met, and the algorithm ends.

[0057] The specific process of step (2) is briefly described below:

[0058] The minimum covering circle of three points can be either the circumscribed circle of the three points or the circle with the line connecting two of the points as its diameter.

[0059] (2.1) Let us first discuss the case where all three points are on the circumference of a circle and find the radius and coordinates of the circle center.

[0060]

[0061]

[0062]

[0063] Assume that the radius of the circle is r, which can be expressed as:

[0064]

[0065] Here, eps is set to 10-10 to prevent the three points from being on a straight line.

[0066] a1=2(x A -x B )

[0067] b1=2(y A -y B )

[0068] a2=2(x B -x C )

[0069] b2=2(y B -y C )

[0070] c1=x A 2 -x B 2 +y A 2 -y B 2

[0071] c2=x B 2 -x C 2 +y B 2 -y C 2

[0072] Now calculate the coordinates (x, y) of the center of the circle:

[0073]

[0074]

[0075] (2.2) Now let's consider a circle whose diameter is the line connecting two points. The necessary and sufficient condition for it to be a minimum point set covering circle is that the third point is inside the circle. Otherwise, the circle cannot be considered a covering circle, and there is no need to compare it with the circumscribed circle of the three points to see if it is a minimum covering circle. Let the midpoint of points A and B be centerPoint(centerPointx, centerPointy). Now find its radius and center coordinates:

[0076]

[0077]

[0078]

[0079]

[0080] Now let’s determine whether the third point is inside the circle:

[0081] If d≤r: then output r and centerPoint as the radius and center

[0082] (2.3) Compare the radius of the circumscribed circle and the radius of the covering circle whose diameter is the line connecting two points in the circle, and take the one with the smallest radius as the minimum covering circle.

[0083] The specific process of step (3) is briefly described below:

[0084] (3.1) After finding the minimum covering circle of points ABC, find the point D farthest from the center of the current minimum covering circle by traversing all points in the point set. If point D is inside the circle, the algorithm ends, the minimum circle containing the point set is found, and the equation of the center radius is output; if point D is not inside the circle, then jump to step (4).

[0085] The specific process of step (4) is briefly described below:

[0086] (4.1) Now we need to find the minimum covering circle of three of the four points ABCD in the previous step, so that it contains these four points and has the smallest radius. By permuting and combining the four points ABCD, we can get three circles ({A,B,C} can no longer form a minimum covering circle) formed by three groups of points on the circumference, namely {A,B,D}, {A,C,D}, {B,C,D}. At this time, we jump to step (2.3) and find the minimum circle radius of each group corresponding to the circle. Compare the three radii and select the circle equation corresponding to the minimum radius and output it. After completing (2.3), we continue to fall into the judgment process of step (3) and iterate until the requirements in step (3) are met. The algorithm ends. At this point, we have obtained the center radius parameter and corresponding equation of the final minimum covering circle containing the entire point set.

[0087] Furthermore, the real-time position of the UAV is obtained by the EKF algorithm to control the UAV to fly to the target point coordinates, including:

[0088] According to the distance information between the tags, the relationship between the known coordinate points and the unknown coordinate points of the tags is constructed;

[0089] Solving the relationship using an EKF algorithm to obtain the coordinates of the unknown coordinate point;

[0090] Get the real-time location coordinates of the drone based on the coordinates of the tag;

[0091] Following the real-time position coordinates of the drone and the target point coordinates, the drone is controlled to fly to the target point coordinates.

[0092] The number of known coordinate nodes measured by the tag is used for judgment. When the number of known coordinate nodes is greater than or equal to 3, the following high-precision positioning algorithm using the adjacent tag distance model can be used:

[0093] like Figure 4 As shown, suppose there are n (n≥3) known nodes, and the node coordinates are (x i ,y i )(i=1,2,…,n), the unknown node coordinates are (x, y), the measured distances between the unknown node and each known node are di(i=1,2,…,n), (1) the distance equation between the known node and the unknown node is established, and the calculation formula of the distance equation (the relationship between the label known coordinate point and the unknown coordinate point) is:

[0094]

[0095] Among them, x and y are the horizontal coordinate value of the unknown coordinate point and the vertical coordinate value of the unknown node, di is the distance from the i-th known node to the unknown node, xi and yi are the horizontal coordinate value and vertical coordinate value of the i-th known node respectively. Figure 4 The r in.

[0096] At this point, we have obtained the radio ranging information of the unknown node and each known node. However, since each set of ranging values ​​contains a large amount of noise, the positioning result directly calculated by least squares will produce a large error. Therefore, we use it as the observation equation of the extended Kalman filter in the subsequent high-precision positioning algorithm.

[0097] The Extended Kalman Filter (EKF) is an extension of the standard Kalman Filter (KF) for nonlinear scenarios. It is a highly efficient recursive filter (autoregressive filter). The basic concept of the EKF is to linearize a nonlinear system using a Taylor series expansion and then filter the signal using the Kalman filter framework. Therefore, it is a suboptimal filter. Because the state transition equations of the CTRV model are nonlinear, and the ranging observation equations of UWB radio signals are also nonlinear, the EKF algorithm is required to achieve high-precision positioning.

[0098] The EKF steps are as follows:

[0099] Prediction step:

[0100] X′ k =f( <X k-1 >)

[0101]

[0102] Get the Kalman gain:

[0103]

[0104]

[0105] Update step:

[0106] <X k > = X′ k +K′ k [Z k -h(X′ k )]

[0107] P k =[1-K′ k H k ]P′ k

[0108] where X′ k represents the prior prediction value of the state, X k Represents the posterior correction value of the state, P′ k represents the prior estimated covariance at time k, P k represents the posterior covariance at time k, K′ k represents the Kalman gain at time k, Q and R represent the variance matrices of process transfer noise and measurement noise respectively, f(·) represents the state transfer equation, h(·) represents the observation equation, and the Jacobian matrix F k-1 and H k They are as follows:

[0109]

[0110] The variance of the measurement noise can be written as:

[0111]

[0112] in They respectively represent the ranging variances between a base station in the observation group and the node to be located.

[0113] For the target to be located, we can use the CTRV model to model and estimate its motion process, and its state variables can be set as:

[0114]

[0115] Its state transfer equation can be described as:

[0116]

[0117] Where subscript k represents the kth moment, v(t) = v k is a constant, and the angular velocity ω is also a constant, and both are independent of time.

[0118] Since the state transfer equation is nonlinear, we can obtain the Jacobian matrix J by taking the partial derivative of each state variable (x, y, v, ω, θ) of the state equation. F :

[0119]

[0120] Where j1 and j2 are:

[0121]

[0122]

[0123] For the observation matrix, if it is nonlinear, we can also use the partial derivative method to calculate the Jacobian matrix J H , for the TOA ranging model, J at this time H It can be expressed as:

[0124]

[0125] For the CTRV mathematical model, the assumed linear acceleration a a and yaw angular acceleration a ω is a constant. However, in the actual motion process, due to the existence of various factors, noise will be added to the state transfer equation, and the noise of the model mainly comes from a a and a ω Under the line-of-sight condition, we can assume that the model noise conforms to the Gaussian distribution, that is, At the same time, the prediction process noise μ of these two acceleration metrics for the state quantities (x, y, v, θ, ω) is:

[0126]

[0127] At this time, the variance of the state transition noise is:

[0128]

[0129] Expand the above formula and substitute it into:

[0130]

[0131] We can then enter the parameters calculated from the above derivation into the EKF equation and iterate through each round to obtain the optimal estimate at each moment. This completes the high-precision positioning process for the node to be located. Once the node's tag coordinates are calculated, its node status is updated to a known coordinate node, and the number of known nodes surrounding the unknown node is refreshed. This means that for all remaining unknown nodes that have measured distances to the tag, the number of known tags measured increases by 1.

[0132] The above process is repeated until the coordinates of all tags with three different tag distances are calculated.

[0133] 4.2 When the number of known coordinate nodes is 2, if the traditional geometric solution method is directly used to solve the equation of the circle to calculate the intersection of the two circles, it will be impossible to determine which intersection it is. However, in the EKF equation, corresponding to the classical method of "solving the equation of the circle to calculate the intersection of the two circles", we can understand it as obtaining two likelihood regions in the observation equation. However, since we can obtain the prior state value of the next moment through the prediction equation in a set of state values ​​at the previous moment, which contains prior position, speed and other information, through the prior information and the two observed likelihood regions, we can eventually guide EKF to determine the correct high likelihood region and achieve correct navigation positioning to improve the robustness of the algorithm. It is worth noting that due to the reduction of the observation dimension at this time, the observation function, the Jacobian matrix of the observation function, and the observation noise covariance h(·), H k , R need to be adjusted in dimension, the principle is the same as above.

[0134] 4.3 When the number of known coordinates is 1, draw a dotted circle with the known coordinate as the center and the measured distance as the radius. Use the intersection of the extension line from other tags to the known coordinate node and the circle as the boundary point. Make the arc part away from other tags a solid line to represent the possible coordinate position of the unknown tag.

[0135] Specifically, this embodiment first models the motion state of the drone, and predicts the drone state at the next moment based on the drone flight state output at the previous moment and the CTRV modeling analysis (corresponding to the EKF prediction step). The EKF obtains the Kalman gain step mainly to analyze the relationship between the modeling prediction noise and the measurement noise to determine its respective trust levels in the modeling and sensor ranging values ​​(mainly a concept of multi-source fusion). The update step corrects the sensor ranging value based on the gain and the state obtained by the previous prediction step only through modeling prediction to obtain the final filtered positioning result.

[0136] The Jacobian matrix is ​​solved in this embodiment because both the state transfer equation and the ranging (observation) equation are nonlinear, and the Jacobian matrix needs to be solved before it can be introduced into the Kalman filter.

[0137] Modeling is necessary in this embodiment because the EKF algorithm is a multi-sensor information fusion algorithm. It is necessary to model and predict the object's motion state to make a priori estimates, and then use the ranging module to perform a posteriori corrections to obtain the filtered results.

[0138] Specifically, the EKF algorithm inputs the filtered state at the previous moment and the ranging information at the current moment. Through modeling and analysis, the filtered state at the previous moment is used to predict the prior state at the current moment. The ranging information is then used to correct the prior state to obtain the filtered state (i.e., the posterior state) at the current moment. The output is the filtered state at the current moment (the state of the CTRV model corresponds to parameters such as coordinates, velocity, heading angle, and turn rate). This output is then fed into the next moment for iteration.

[0139] The beneficial effects of the present invention are as follows:

[0140] The present invention provides a method for realizing autonomous drone following. The present invention converts the distance relationship between the drone base station and the tag into a point set minimum circle problem, finds the optimal target point of the drone, calculates the coordinates of the optimal target point of the drone, and controls the drone to fly to the coordinates of the optimal target point, thereby realizing autonomous following of the drone base station.

[0141] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0142] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for realizing autonomous drone following, characterized in that: include: Use time of flight or two-way ranging to obtain ranging information between a tag with a known location and a tag with an unknown location; Based on the ranging information, the location information of the unknown location tags is continuously calculated using the EKF algorithm until the location information of all tags is obtained; Based on the position information of all the tags, the optimal following position of the drone and the tags is solved using the point increment method; Obtaining the coordinates of the target point followed by the drone based on the coordinates of the tag and the optimal following position of the drone and the tag; Obtain the real-time position of the UAV through the EKF algorithm to control the UAV to fly to the coordinates of the following target point; Wherein, based on the position information of all the tags, solving the optimal following position of the drone and the tags using the point increment method includes: Obtaining the relative position vector of the drone and the tag parallel to the xy plane and the relative position vector parallel to the z axis; eliminating the relative position vector parallel to the z axis; The tags to be located are all on the ground, and their relative position vectors for distance measurement are all parallel to the xy plane. However, since the drone is flying in the air, the distance measurement information between the drone and the tag includes distance measurement information on the z-axis. The distance between the drone and the tag is decomposed into a relative position vector parallel to the xy plane and a relative position vector parallel to the z-axis. The projection of the drone's optimal following position on the xy plane is obtained, and only the relative position vector parallel to the xy plane is retained in the distance measurement equation. Obtaining the coordinates of the target point to be followed by the drone based on the coordinates of the tags and the optimal following positions of the drone and the tags, including: constructing a first preset circle, where the first preset circle is a minimum circle containing the coordinate points of all the tags; constructing a second preset circle, where the circumference of the second preset circle passes through the coordinate points corresponding to any three tags, or only passes through two of the three coordinate points, with the other coordinate point located within the circle; the any three tags are a first tag, a second tag, and a third tag; Obtain the coordinate point of the tag corresponding to the maximum distance between the coordinate points of all tags and the center of the second preset circle to obtain the first preset coordinate point; determine whether the first preset coordinate point is on the circumference of the second preset circle or within the circle; if so, obtain the corresponding equation and center radius parameter formula of the second preset circle and obtain the coordinates of the target point followed by the drone according to the corresponding equation and corresponding center radius parameter of the second preset circle; if not, construct a third preset circle until the third preset circle meets the judgment condition of the current preset circle and obtain the equation and corresponding center radius parameter corresponding to the third preset circle and obtain the coordinates of the target point followed by the drone according to the corresponding equation and corresponding center radius parameter of the third preset circle; the third preset circle is any three points of the four points including the coordinate point of the first tag, the coordinate point of the second tag, the coordinate point of the third tag and the first preset coordinate point, and the third preset circle cannot coincide with the second preset circle.

2. The method for realizing autonomous drone following according to claim 1, characterized in that: The obtaining the coordinates of the target point followed by the drone according to the coordinates of the tag and the optimal following position of the drone and the tag includes: Constructing a first preset circle, where the first preset circle is a minimum circle containing the coordinate points of all tags; Constructing a second preset circle, wherein the circumference of the second preset circle passes through the coordinate points corresponding to any three tags, or passes through only two of the three coordinate points, with the other coordinate point located inside the circle; the any three tags are the first tag, the second tag, and the third tag; Obtaining the coordinate point of the tag corresponding to the maximum distance between the coordinate points of all tags and the center of the second preset circle to obtain the first preset coordinate point; Determine whether the first preset coordinate point is on or within the circumference of the second preset circle. If so, obtain the corresponding equation and center radius parameter formula of the second preset circle and obtain the coordinates of the target point followed by the drone according to the corresponding equation and corresponding center radius parameter of the second preset circle. If not, construct a third preset circle until the third preset circle meets the judgment condition of the current preset circle and obtain the equation and corresponding center radius parameter corresponding to the third preset circle and obtain the coordinates of the target point followed by the drone according to the corresponding equation and corresponding center radius parameter of the third preset circle. The third preset circle is any three points of the four points including the coordinate point of the first label, the coordinate point of the second label, the coordinate point of the third label and the first preset coordinate point, and the third preset circle cannot coincide with the second preset circle.

3. The method for realizing autonomous drone following according to claim 2, characterized in that: The second preset circle is a circumscribed circle of the coordinate points of any three tags or a circle whose diameter is a line connecting any two of the coordinate points of the three tags, and the coordinate point of another tag is inside the circle.

4. The method for realizing autonomous drone following according to claim 1, characterized in that: Obtaining the real-time position of the UAV through the EKF algorithm to control the UAV to fly to the target point coordinates includes: According to the distance information between the tags, the relationship between the known coordinate points and the unknown coordinate points of the tags is constructed; Solving the relationship using an EKF algorithm to obtain the coordinates of the unknown coordinate point; Get the real-time location coordinates of the drone based on the coordinates of the tag; The drone is controlled to fly to the target point coordinates according to the real-time position coordinates of the drone and the target point coordinates.

5. The method for realizing autonomous drone following according to claim 4, characterized in that: The relationship between the known coordinate points and the unknown coordinate points of the tag is: ； Among them, x and y are the horizontal coordinate value of the unknown coordinate point and the vertical coordinate value of the unknown node respectively, d i is the distance from the i-th known node to the unknown node, x i and y i are the horizontal and vertical coordinate values ​​of the i-th known node respectively.

Images