Autonomous refueling control method for unmanned aerial vehicle
Through the combination of DGPS sensor and vision sensor, the BP neural network and PID controller, the high-precision tracking of the tanker cone by the unmanned aerial vehicle is realized, solving the problem of position deviation control between the tanker and the oil receiver in high altitude environment, and improving the unmanned aerial vehicle endurance and combat radius.
Patent Information
- Application Number
- CN202411078909.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-07
- Publication Date
- 2024-12-06
AI Technical Summary
Unmanned aerial vehicles have poor endurance in high altitude environments. Traditional refueling technology is affected by satellite signal update frequency and interference, making it difficult to achieve high-precision tracking and positioning, especially under the influence of atmospheric turbulence, which is difficult to control the position deviation between the refueling engine and the oil-receiving engine.
After the initial meeting of the DGPS sensor, the image of the gas tanker cone is collected through the visual sensor, feature extraction and elliptical recognition are performed, combined with the BP neural network and the PID controller, a spatial rectangular coordinate system is established, the position information is updated, and the stable tracking of the gas tanker is achieved. .
It improves the accuracy and stability of the refueling process of unmanned aerial vehicles, enhances the ability to adapt to complex environmental interference, and improves the effective combat radius.
Smart Images

Figure CN119087780A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of autonomous refueling, and in particular relates to an autonomous refueling control method for an unmanned aerial vehicle. Background Art
[0002] With the continuous development of science and technology, unmanned aerial vehicles are widely used in military and civilian fields, especially in many military tasks such as enemy reconnaissance, communication relay and air strike. However, the main factor restricting the development of unmanned aerial vehicles is the poor endurance and the flight radius cannot meet people's use needs. Therefore, the emergence of autonomous aerial refueling technology for unmanned aerial vehicles can effectively solve this problem. According to statistical estimates, the effective combat radius of unmanned aerial vehicles can be increased by 40%~80% through the autonomous refueling technology of unmanned aerial vehicles. Therefore, it is particularly important to vigorously develop autonomous refueling technology for unmanned aerial vehicles. Since the flight altitude of unmanned aerial vehicles is generally 4000m to 10000m, unmanned aerial vehicles in this altitude range are easily disturbed by atmospheric turbulence, making it more difficult to control the position deviation of the refueling machine and the unmanned receiving machine. Therefore, this requires that the unmanned aerial vehicle not only needs to be equipped with a powerful perception system that can efficiently identify the refueling cone to achieve autonomous refueling, but also needs to have an adaptive, highly robust and high-precision control algorithm to cope with complex and uncertain interference in the environment.
[0003] Traditional UAV refueling technology uses DGPS relative positioning technology to track and locate the UAV (refueling aircraft) and the refueling aircraft. Although DGPS relative positioning technology can provide high-precision location information with only a small delay, when the UAV flies too fast, the location information update frequency provided by the satellite may not meet the UAV tracking requirements at high speeds. At the same time, when the satellite signal is lost or the signal is interfered, it will also have a certain impact on the tracking and positioning of the UAV. Summary of the invention
[0004] The purpose of the present invention is to solve the problems raised in the background technology and to provide an autonomous refueling control method for an unmanned aerial vehicle.
[0005] To achieve the above object, the technical solution adopted by the present invention is: The present invention provides an autonomous refueling control method for an unmanned aerial vehicle, comprising: When the receiving aircraft needs to refuel, it will initially obtain the location of the refueling aircraft and fly to the preset location to meet with the refueling aircraft; The receiving aircraft collects images of the drogue of the refueling machine at preset time intervals, and pre-processes each collected image to obtain a characteristic map of the drogue; For each feature map of the cone sleeve, define the edge points, then determine whether any two edge points belong to the same category, group all edge points, and connect the edge points in each group to obtain the corresponding arc segment; Then, all arc segments are grouped, and each group of arc segments is screened to select arc segments that meet the characteristics of the ellipse, and each group of screened arc segments is used to construct a corresponding ellipse; Comparing the obtained ellipses with a preset template image of the fuel filling port of the drogue, screening out ellipses with a similarity greater than a first threshold, and defining the screened out ellipses of this type as candidate ellipses; Compare the similarity of each candidate ellipse in the last acquired image with each candidate ellipse in the current acquired image, and use the areas where the two candidate ellipses with the highest similarity are located as the areas before and after the refueling port of the drogue moves respectively; In each interval time period, a spatial rectangular coordinate system is established for the refueling port of the drogue sleeve and the refueling receiving machine, and the first position information of the refueling port of the drogue sleeve relative to the refueling receiving machine is obtained according to the conversion relationship between the two spatial rectangular coordinate systems, and the first position information is updated once every preset time period; In each interval time period, the first position information obtained is subtracted from the current position of the receiving aircraft, and then input into the BP neural network and PID controller respectively. The BP neural network outputs the adjustment parameter as the adjustment parameter of the PID controller. After the PID controller outputs the control signal, it outputs the position signal to the receiving aircraft through the filter. The receiving aircraft updates and adjusts the current position according to the position signal. This process is repeated to achieve stable tracking of the refueling port of the cone sleeve by the receiving aircraft.
[0006] Preferably, the receiving aircraft is provided with a DGPS sensor for obtaining the location of the refueling machine; The receiving aircraft is also provided with at least one visual sensor, and the receiving aircraft collects images through the visual sensor.
[0007] Preferably, the process of preprocessing each collected image includes: Delete error frame information; Adjust the brightness of each collected image; Perform grayscale processing on each collected image.
[0008] Preferably, judging whether any two edge points belong to the same class includes: The similarity of the two edge points is calculated according to the grayscale value and gradient amplitude of the two edge points, and the specific formula is as follows: ; in, Represents two edge points and The similarity between Represents edge points The gray value of Represents edge points The gray value of Represents edge points The gradient amplitude of Represents edge points The gradient amplitude of When the similarity between two edge points exceeds a second threshold, the two edge points belong to the same class.
[0009] Preferably, when the arc segments are grouped, the arc segments are grouped according to their curvatures, and the arc segments whose curvature differences are less than a fourth threshold are grouped into the same group.
[0010] Preferably, when each ellipse is compared with a preset template image of the fuel filler port of the cone sleeve, the size of each ellipse is compared with the size of the fuel filler port in the template image and the difference is taken. When the difference is less than a preset third threshold, the ellipse is considered to be an ellipse with a similarity greater than the first threshold.
[0011] Preferably, when comparing the similarity of each candidate ellipse in the last acquired image with each candidate ellipse in the currently acquired image, the similarity of each candidate ellipse in the last acquired image is not repeatedly compared with each candidate ellipse in the currently acquired image, and the similarity comparison is performed through a twin network.
[0012] Preferably, in each interval time period, a spatial rectangular coordinate system is established for the refueling port of the drogue sleeve and the refueling machine, and the first position information of the refueling port of the drogue sleeve relative to the refueling machine is obtained according to the conversion relationship between the two spatial rectangular coordinate systems, and the first position information is updated once every preset time period, including: In each interval time period, a first spatial rectangular coordinate system is established for the refueling port of the drogue sleeve, and a second spatial rectangular coordinate system is established for the receiving aircraft. The coordinate of any point in the refueling port of the drogue sleeve in the first spatial rectangular coordinate system is P , the point P is projected onto the image collected by the receiving aircraft and becomes point Q, and the coordinates of point Q in the second spatial rectangular coordinate system are , the transformation relationship between point P and point Q satisfies: (1); in, represents a rotation transformation, represents a translation transformation; and, (2); ; in, Indicates the pitch angle of the drogue refueling port relative to the receiving aircraft. Indicates the yaw angle of the drogue refueling port relative to the receiving aircraft. Indicates the roll angle of the refueling port of the cone sleeve relative to the receiving machine. represents the translation along the x-axis direction of the second spatial rectangular coordinate, represents the translation along the y-axis direction of the second spatial rectangular coordinate, represents the translation amount along the z-axis direction of the second spatial rectangular coordinate; The rotation transformation in the current time interval and all previous time intervals is obtained through the conversion relationship and translation transformations The value of , and then the rotation transformation of all the interval time periods obtained Perform weighted averaging to obtain the average rotation transformation in each interval time period The rotation transformation value of the drogue of the refueling machine relative to the receiving machine; The average rotation transform Substituting into formula (2), we get: ; Then we can solve the problem , and value, and , and The values are used together as the first position information, Indicates the average pitch angle of the drogue refueling port relative to the receiving aircraft. Indicates the average yaw angle of the drogue refueling port relative to the receiving aircraft. Indicates the average roll angle of the drogue's refueling port relative to the receiving aircraft.
[0013] Compared with the prior art, the present invention has the following beneficial effects: The present unmanned aerial vehicle autonomous refueling control method firstly uses a DGPS sensor to achieve a preliminary rendezvous between the receiving aircraft and the refueling aircraft, and then uses a visual sensor to collect an image of the cone sleeve of the refueling aircraft, and processes the image to obtain an elliptical area where the refueling port is located in the image, and then establishes a spatial rectangular coordinate system for the refueling port of the cone sleeve and the receiving aircraft respectively, and obtains the first position information of the refueling port of the cone sleeve relative to the receiving aircraft according to the conversion relationship between the two spatial rectangular coordinate systems, and then uses a BP neural network combined with a PID control algorithm to continuously adjust the position of the receiving aircraft, thereby achieving stable tracking of the refueling port of the cone sleeve by the receiving aircraft. Compared with the prior art, the tracking of this method is more accurate and stable. BRIEF DESCRIPTION OF THE DRAWINGS
[0014] Figure 1 It is a flow chart of the autonomous refueling control method of the unmanned aerial vehicle of the present invention; Figure 2 A schematic diagram of a first space rectangular coordinate system and a second space rectangular coordinate system established for the present invention; Figure 3 It is a flow chart of PID control of the present invention. DETAILED DESCRIPTION
[0015] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0016] It should be noted that when a component is referred to as being "connected" to another component, it may be directly connected to the other component or there may be an intermediate component. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application.
[0017] like Figure 1-Figure 3 As shown, a method for controlling autonomous refueling of an unmanned aerial vehicle comprises: It should be noted that the cone sleeve of the fuel dispenser may be affected by atmospheric turbulence and the tail vortex of the fuel dispenser, causing its position to shift. Therefore, the cone sleeve of the fuel dispenser is in motion. This patent first makes a preliminary rendezvous between the receiving machine and the fuel dispenser, and then allows the receiving machine to maintain stable tracking of the refueling port of the cone sleeve (when refueling, the refueling port of the cone sleeve refuels the fuel tank of the receiving machine), thereby achieving docking between the receiving machine and the fuel dispenser, which is convenient for subsequent refueling.
[0018] S1. When the receiving aircraft needs to refuel, it will initially obtain the location of the refueling aircraft and fly to the preset location to meet with the refueling aircraft; The receiving aircraft is equipped with a DGPS sensor for preliminarily acquiring the location of the refueling aircraft. After knowing the preliminary location of the refueling aircraft, the receiving aircraft flies to a preset location to meet with the refueling aircraft. For example, the preset location can be within a preset range of 10 meters from the refueling aircraft to achieve rendezvous with the refueling aircraft.
[0019] S2, the receiving aircraft collects images of the drogue of the refueling machine at preset time intervals, and pre-processes each collected image to obtain a characteristic map of the drogue; In this embodiment, the preset time interval is the time period from when the receiving machine acquires the image of the drogue of the refueling machine to when the first position information of the refueling port of the drogue relative to the receiving machine is obtained; During the process of refueling the receiving aircraft, the refueling machine uses the refueling port of the cone sleeve to access the oil storage tank of the receiving aircraft. Therefore, the receiving aircraft acquires images of the cone sleeve of the refueling machine (collects multiple images at one time). The receiving aircraft is equipped with at least one visual sensor, which acquires images of the cone sleeve of the refueling machine through the visual sensor. The process of preprocessing the collected images includes: Delete error frame information (such as blurred images, etc.); Adjust the brightness of each collected image (for example, brighten underexposed images and darken overexposed images); Perform grayscale processing on each collected image (through grayscale processing, reduce the interference of different color changes on the refueling cone sleeve and improve tracking stability).
[0020] S3, for each image collected at one time, for each feature map of the cone sleeve, define edge points (define edge points for the entire feature map, that is, define certain points in the entire feature map as edge points), then determine whether any two edge points belong to the same category, group all edge points, and connect the edge points in each group to obtain corresponding arc segments; Among them, the edge points are composed of position, gray value and gradient amplitude; Determine whether any two edge points belong to the same class, including: The similarity of the two edge points is calculated according to the gray value and gradient amplitude of the two edge points, and the specific formula is as follows: ; in, Represents two edge points and The similarity between Represents edge points The gray value of Represents edge points The gray value of Represents edge points The gradient amplitude of Represents edge points The gradient amplitude of When the similarity between two edge points exceeds the second threshold, the two edge points belong to the same class, and the edge points in each feature map are grouped, and then the edge points in each group are connected into arc segments.
[0021] S4, for each image collected at one time, grouping all arc segments, screening each group of arc segments, screening out arc segments that meet the ellipse characteristics, and using each group of screened arc segments to construct a corresponding ellipse; When the arc segments are grouped, they are grouped according to the curvature of the arc segments, and the arc segments whose curvature difference is less than the fourth threshold are divided into the same group (or grouped by using a similarity feature classification algorithm, such as Euclidean distance, Manhattan distance, cosine similarity, etc.); each group of arc segments is screened, and when the arc segments that meet the ellipse characteristics are screened, the first screening is performed by the smoothness of the arc, and most of the adhesion arcs can be eliminated by the smoothness; the second screening is also performed by the concavity and convexity of the arc, and each group of arc segments after the first screening Calculate the center position of the ellipse formed by the group of arc segments, and calculate whether the distances from each group of arc segments after screening to the center position are the same. If they are the same, it proves that these arc segments belong to the same ellipse (eliminate the different arc segments), and then construct the corresponding ellipse for each group of arc segments after the second screening, and then perform smoothness processing on these elliptical areas to enhance the curvature and smoothness of the edge lines of the elliptical areas, so as to obtain the elliptical areas in the image; then perform black and white grayscale processing on the obtained elliptical areas to highlight the edge lines of the elliptical areas, making the characteristics of the arcs more obvious.
[0022] S5, comparing the obtained ellipses with a preset template image of the fuel filling port of the drogue, screening out ellipses with a similarity greater than a first threshold, and defining the screened out ellipses of this type as candidate ellipses; Among them, the template image of the fuel filler port in the cone sleeve is an image of the fuel filler port determined by a preset size. When each ellipse is compared with the preset template image of the fuel filler port of the cone sleeve, the size of each ellipse is compared and subtracted from the size of the fuel filler port in the template image (for example, the radius, circumference or area of each ellipse can be respectively compared and subtracted with the radius, circumference or area of the fuel filler port in the template image, or an ellipse that meets all three conditions of radius, circumference and area can be regarded as an ellipse with a similarity greater than a first threshold, such as first judging whether the radius meets the requirements, then judging whether the circumference meets the requirements, and finally judging whether the area meets the requirements). When the difference is less than a preset third threshold, the ellipse is considered to be an ellipse with a similarity greater than the first threshold.
[0023] S6, comparing the similarity of each candidate ellipse in the last acquired image with each candidate ellipse in the current acquired image, and using the regions where the two candidate ellipses with the highest similarity are located as the regions before and after the refueling port of the drogue moves; When comparing the similarity of each candidate ellipse in the last acquired image with each candidate ellipse in the current acquired image, the candidate ellipses in the last acquired image are not repeatedly compared with the candidate ellipses in the current acquired image (that is, arbitrarily select one candidate ellipse from each candidate ellipse in the last acquired image and arbitrarily select one candidate ellipse from each candidate ellipse in the current acquired image for similarity comparison, and the compared two candidate ellipses are not repeatedly compared), and the similarity comparison is performed through a twin network, wherein two convolutional neural networks are used as two branches in the twin network respectively. When comparing, the two compared ellipses are respectively input into the two convolutional neural networks, and the corresponding feature maps are output. Then, the two feature maps are compared for similarity (the Euclidean distance can be calculated), and the areas where the two candidate ellipses with the highest similarity are located are respectively used as the areas before and after the refueling port of the cone sleeve moves (the ellipse in the last acquired image is used as the area before the refueling port moves, and the ellipse in the current acquired image is used as the area after the refueling port moves).
[0024] S7, in each interval time period, respectively establish a spatial rectangular coordinate system for the refueling port of the drogue sleeve and the refueling machine, and obtain the first position information of the refueling port of the drogue sleeve relative to the refueling machine according to the conversion relationship between the two spatial rectangular coordinate systems, and the first position information is updated once every preset time period; Specifically: in each interval time period, establish the first spatial rectangular coordinate system for the refueling port of the cone sleeve (in Figure 2 The first space rectangular coordinate system is It means that if the first spatial rectangular coordinate system is established with the center point of the fuel filler port as the origin, Figure 2 The cone in the middle is the cone sleeve. A second rectangular coordinate system is established for the receiving machine (in Figure 2 The second rectangular coordinate system in space is It means that if the second spatial rectangular coordinate system is established with the center point of the image of the visual sensor on the receiving aircraft as the origin, Figure 2 The rectangle in the figure is the image of the visual sensor). The coordinates of any point in the refueling port of the cone sleeve in the first spatial rectangular coordinate system are P , the point P is projected onto the image collected by the receiving aircraft and becomes point Q (that is, point P is obtained by collecting the image of the visual sensor to obtain point Q on the image, that is, point Q is the point after the imaging of point P), and the coordinates of point Q in the second spatial rectangular coordinate system are , the transformation relationship between point P and point Q satisfies: (1); in, represents a rotation transformation, represents a translation transformation; and, (2); ; in, Indicates the pitch angle of the drogue refueling port relative to the receiving aircraft. Indicates the yaw angle of the drogue refueling port relative to the receiving aircraft. Indicates the roll angle of the refueling port of the cone sleeve relative to the receiving machine. Represents the rectangular coordinates along the second space The amount of translation in the axis direction, Represents the rectangular coordinates along the second space The amount of translation in the axis direction, Represents the rectangular coordinates along the second space The amount of translation in the axis direction; The rotation transformation in the current time interval and all previous time intervals is obtained through the conversion relationship and translation transformations The value of , and then the rotation transformation of all the interval time periods obtained Perform weighted averaging to obtain the average rotation transformation in each interval time period The rotation transformation value of the drogue of the refueling machine relative to the receiving machine; The average rotation transform Substituting into formula (2), we get: ; Expand the above formula into three equations, and then solve them to get , and value, and , and The values are used together as the first position information, Indicates the average pitch angle of the drogue refueling port relative to the receiving aircraft. Indicates the average yaw angle of the drogue refueling port relative to the receiving aircraft. Indicates the average roll angle of the drogue's refueling port relative to the receiving aircraft.
[0025] Since the first position information is transformed according to the average rotation Calculated, the average rotation transformation The first position information is continuously updated as the interval time period increases, so the first position information is updated once every preset time period (the preset time period is the time period from when the receiving aircraft collects the image of the fuel dispenser cone sleeve to when the first position information of the cone sleeve refueling port relative to the receiving aircraft is obtained).
[0026] S8. In each interval time period, the obtained first position information is subtracted from the current position of the receiving aircraft, and then input into the BP neural network and the PID controller respectively. The BP neural network outputs the adjustment parameter as the adjustment parameter of the PID controller. After the PID controller outputs the control signal, it outputs the position signal to the receiving aircraft through the filter. The receiving aircraft updates and adjusts the current position according to the position signal. This process is repeated to achieve stable tracking of the fueling port of the drogue by the receiving aircraft. Among them, Figure 3 As shown in the figure, the first position information obtained in the first interval is subtracted from the current position of the receiving machine and then input into the BP neural network and PID controller respectively. The BP neural network outputs the adjustment parameter (proportional parameter , integration parameters and the differential parameters ) to the PID controller as the adjustment parameter of the PID controller, and then the PID controller outputs a control signal, and the controller signal then outputs a position signal to the receiving aircraft through a filter, and the receiving aircraft updates and adjusts its current position according to the position signal (for example, it can specifically adjust the receiving aircraft's motor to drive the receiving aircraft to move according to the position signal), and then the first position information obtained within the second interval is subtracted from the updated current position of the receiving aircraft, and then respectively input into the BP neural network and the PID controller, and the cycle is repeated to achieve stable tracking of the refueling port of the cone sleeve by the receiving aircraft, thereby achieving docking between the receiving aircraft and the refueling machine, which is convenient for subsequent refueling.
[0027] The input layer of the BP neural network consists of two neurons, and the hidden layer consists of three neurons. The output function of each neuron is different, corresponding to the proportional parameters. , integration parameters and the differential parameters .
[0028] The present unmanned aerial vehicle autonomous refueling control method firstly uses a DGPS sensor to achieve a preliminary rendezvous between the receiving aircraft and the refueling aircraft, and then uses a visual sensor to collect an image of the cone sleeve of the refueling aircraft, and processes the image to obtain an elliptical area where the refueling port is located in the image, and then establishes a spatial rectangular coordinate system for the refueling port of the cone sleeve and the receiving aircraft respectively, and obtains the first position information of the refueling port of the cone sleeve relative to the receiving aircraft according to the conversion relationship between the two spatial rectangular coordinate systems, and then uses a BP neural network combined with a PID control algorithm to continuously adjust the position of the receiving aircraft, thereby achieving stable tracking of the refueling port of the cone sleeve by the receiving aircraft. Compared with the prior art, the tracking of this method is more accurate and stable.
[0029] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0030] The above-described embodiments only express the more specific and detailed embodiments described in this application, but they cannot be understood as limiting the scope of the patent application. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of this application, which all belong to the protection scope of this application. Therefore, the protection scope of the patent application shall be based on the attached claims.
Claims
1. A method for controlling autonomous refueling of an unmanned aerial vehicle, characterized in that: The unmanned aerial vehicle autonomous refueling control method comprises: When the receiving aircraft needs to refuel, it will initially obtain the location of the refueling aircraft and fly to the preset location to meet with the refueling aircraft; The receiving aircraft collects images of the drogue of the refueling machine at preset time intervals, and pre-processes each collected image to obtain a characteristic map of the drogue; For each feature map of the cone sleeve, define the edge points, then determine whether any two edge points belong to the same category, group all edge points, and connect the edge points in each group to obtain the corresponding arc segment; Then, all arc segments are grouped, and each group of arc segments is screened to select arc segments that meet the characteristics of the ellipse, and each group of screened arc segments is used to construct a corresponding ellipse; Comparing the obtained ellipses with a preset template image of the fuel filling port of the drogue, screening out ellipses with a similarity greater than a first threshold, and defining the screened out ellipses of this type as candidate ellipses; Compare the similarity of each candidate ellipse in the last acquired image with each candidate ellipse in the current acquired image, and use the areas where the two candidate ellipses with the highest similarity are located as the areas before and after the refueling port of the drogue moves respectively; In each interval time period, a spatial rectangular coordinate system is established for the refueling port of the drogue sleeve and the refueling receiving machine, and the first position information of the refueling port of the drogue sleeve relative to the refueling receiving machine is obtained according to the conversion relationship between the two spatial rectangular coordinate systems, and the first position information is updated once every preset time period; In each interval time period, the first position information obtained is subtracted from the current position of the receiving aircraft, and then input into the BP neural network and PID controller respectively. The BP neural network outputs the adjustment parameter as the adjustment parameter of the PID controller. After the PID controller outputs the control signal, it outputs the position signal to the receiving aircraft through the filter. The receiving aircraft updates and adjusts the current position according to the position signal. This process is repeated to achieve stable tracking of the refueling port of the cone sleeve by the receiving aircraft.
2. The autonomous refueling control method for an unmanned aerial vehicle as claimed in claim 1, characterized in that: The receiving aircraft is provided with a DGPS sensor for obtaining the location of the refueling machine; The receiving aircraft is also provided with at least one visual sensor, and the receiving aircraft collects images through the visual sensor.
3. The autonomous refueling control method for an unmanned aerial vehicle as claimed in claim 1, characterized in that: The process of preprocessing the collected images includes: Delete error frame information; Adjust the brightness of each collected image; Perform grayscale processing on each collected image.
4. The autonomous refueling control method for an unmanned aerial vehicle as claimed in claim 1, characterized in that: Determine whether any two edge points belong to the same class, including: The similarity of the two edge points is calculated according to the gray value and gradient amplitude of the two edge points, and the specific formula is as follows: ; in, Represents two edge points and The similarity between Represents edge points The gray value of Represents edge points The gray value of Represents edge points The gradient amplitude of Represents edge points The gradient amplitude of When the similarity between two edge points exceeds a second threshold, the two edge points belong to the same class.
5. The autonomous refueling control method for an unmanned aerial vehicle as claimed in claim 1, characterized in that: When the arc segments are grouped, the arc segments are grouped according to their curvatures, and the arc segments whose curvature differences are less than a fourth threshold are grouped into the same group.
6. The autonomous refueling control method for an unmanned aerial vehicle as claimed in claim 1, characterized in that: When each ellipse is compared with a preset template image of the fuel filler port of the cone sleeve, the size of each ellipse is compared with the size of the fuel filler port in the template image and a difference is made. When the difference is less than a preset third threshold, the ellipse is considered to be an ellipse with a similarity greater than the first threshold.
7. The autonomous refueling control method for an unmanned aerial vehicle as claimed in claim 1, characterized in that: When comparing the similarity of each candidate ellipse in the last acquired image with each candidate ellipse in the current acquired image, the similarity of each candidate ellipse in the last acquired image is not repeated and the similarity comparison is performed through the twin network.
8. The autonomous refueling control method for an unmanned aerial vehicle as claimed in claim 1, characterized in that: In each interval time period, a spatial rectangular coordinate system is established for the refueling port of the drogue sleeve and the refueling receiving machine, and first position information of the refueling port of the drogue sleeve relative to the refueling receiving machine is obtained according to the conversion relationship between the two spatial rectangular coordinate systems, and the first position information is updated once every preset time period, including: In each interval time period, a first spatial rectangular coordinate system is established for the refueling port of the drogue sleeve, and a second spatial rectangular coordinate system is established for the receiving aircraft. The coordinate of any point in the refueling port of the drogue sleeve in the first spatial rectangular coordinate system is P , the point P is projected onto the image collected by the receiving aircraft and becomes point Q, and the coordinates of point Q in the second spatial rectangular coordinate system are , the transformation relationship between point P and point Q satisfies: (1); in, represents a rotation transformation, represents a translation transformation; and, (2); ; in, Indicates the pitch angle of the drogue refueling port relative to the receiving aircraft. Indicates the yaw angle of the drogue refueling port relative to the receiving aircraft. Indicates the roll angle of the refueling port of the cone sleeve relative to the receiving machine. represents the translation along the x-axis direction of the second spatial rectangular coordinate, represents the translation along the y-axis direction of the second spatial rectangular coordinate, represents the translation amount along the z-axis direction of the second spatial rectangular coordinate; The rotation transformation in the current time interval and all previous time intervals is obtained through the conversion relationship and translation transformations The value of , and then the rotation transformation of all the interval time periods obtained Perform weighted averaging to obtain the average rotation transformation in each interval time period The rotation transformation value of the drogue of the refueling machine relative to the receiving machine; The average rotation transform Substituting into formula (2), we get: ; Then we can solve the problem , and value, and , and The values are used together as the first position information, Indicates the average pitch angle of the drogue refueling port relative to the receiving aircraft. Indicates the average yaw angle of the drogue refueling port relative to the receiving aircraft. Indicates the average roll angle of the drogue's refueling port relative to the receiving aircraft.
