CN113790719B - A UAV inertial/visual landing navigation method based on line features - Google Patents

Abstract

The invention discloses an unmanned aerial vehicle inertial/visual landing navigation method based on line characteristics, which comprises the steps of firstly, collecting images of an airport runway, extracting real-time characteristics of the runway edge, and obtaining a linear equation of the edge and a central line; calculating Plucker coordinates of the boundary line equation through the width of the airport runway and the camera internal reference matrix which are bound in advance; calculating an equation of an infinite distance blanking point and a blanking line by two equidistant parallel lines, solving an attitude transfer matrix between a real-time world coordinate system and a camera coordinate system of the unmanned aerial vehicle by a simultaneous equation set, and solving the attitude, lateral and vertical positions; and taking the position information calculated by the visual landing system as an observed quantity, and fusing the observed quantity with a Kalman filter constructed by the navigation information output by inertial navigation to realize a continuous and autonomous navigation positioning function. The invention solves the problems of accumulated divergence of inertial navigation errors along with time and larger noise of visual navigation resolving results.

Description

A UAV inertial/visual landing navigation method based on line features

Technical Field

The invention belongs to the field of navigation technology, and in particular relates to a drone landing navigation method.

Background Art

The line features used in the posture solution of UAV visual landing are the two side lines of the airport runway, the center line, and the distant hidden line. Unlike point features, line features are not easily affected by factors such as lighting, flight distance, and altitude, and have better robustness. However, since the features of the hidden line are not obvious enough, it is difficult to identify them through feature extraction. A new solution is needed to obtain the equation of the hidden line in the image plane. At the same time, it is necessary to consider that the posture solution results of the visual navigation method are not smooth enough and will jump with the rapid maneuvering of the UAV. A filtering method is needed to fuse the inertial and visual navigation results.

Summary of the invention

The present invention provides an unmanned aerial vehicle inertial/visual landing navigation method based on line features, which solves the problems of cumulative divergence of inertial navigation errors over time and large noise in visual navigation solution results.

A UAV inertial/visual landing navigation method based on line features comprises the following steps:

(1) Airport runway data collection

Establish the airport coordinate system, visual coordinate system, world coordinate system, camera coordinate system, and image coordinate system; collect images of the airport runway, perform real-time feature extraction on the runway sidelines, and obtain the linear equations of the sidelines and centerlines;

(2) Plücker coordinate representation

The Plücker coordinates of the edge equation are calculated by using the pre-bound airport runway width and the camera internal parameter matrix;

(3) Visual measurement pose calculation

After obtaining the Plücker coordinates of the runway edge, the equations of the vanishing point and the vanishing line at infinity are calculated by two equidistant parallel lines. The attitude transfer matrix C _w c between the real-time world coordinate system and the camera coordinate system of the UAV is solved by the simultaneous equations, and the attitude and lateral and vertical positions are solved.

(4) Combined navigation based on inertial/visual fusion

The position information calculated by the visual landing system is used as the observation quantity, and the Kalman filter is constructed to fuse it with the navigation information output by the inertial navigation to realize continuous and autonomous navigation and positioning functions.

Furthermore, in step (1), the airport coordinate system, the visual coordinate system, the world coordinate system, the camera coordinate system, and the image coordinate system are established, including:

The airport coordinate system is denoted as the a system; the intersection of the runway landing end start line and the runway center line is the origin _oa ; the axle is along the runway center line, and the forward direction is positive; the _{ya axis is perpendicular to the runway plane, and the upward direction is positive; the z a axis} coincides with the runway start line, and the right direction is positive; _oa x _a y _a z _a constitutes a right-handed coordinate system; the coordinates of a point in the airport coordinate system are expressed as (x _a , y _a , z _a );

Visual coordinate system, denoted as v system; the image principal point of the optical system is taken as the origin o _v ; the x _v axis is parallel to the optical axis, and the forward direction is positive; the y _v axis is parallel to the horizontal axis of the imaging plane coordinate system, and the upward direction is positive; the z _v axis, the x _v axis and the y _v axis form a right-handed coordinate system, and the right direction is positive;

The world coordinate system is recorded as the w system; the intersection of the runway landing end aiming point starting line and the runway centerline is taken as the origin o _w ; the x _w axis coincides with the runway starting line, and the right direction is positive; the y _w axis is perpendicular to the runway plane, and the downward direction is positive; the z _w axis is along the runway centerline, and the forward direction is positive; o _w x _w y _w z _w constitutes a right-handed coordinate system; the coordinates of a point in the world coordinate system are expressed as (x _w ,y _w ,z _w );

The camera coordinate system is denoted as the c system; the image principal point of the optical system is taken as the origin o _c ; when observing the optical system directly, the x _c axis is parallel to the horizontal axis of the imaging plane coordinate system, and the left direction is positive; the y _c axis is parallel to the vertical axis of the imaging plane coordinate system, and the downward direction is positive; the z _c axis points to the observer and forms a right-handed coordinate system with the x _c axis and the y _c axis;

The image coordinate system is denoted as the i-system; the image coordinate system is established in the plane where the camera's photosensitive surface is located, with the upper left corner of the image as the origin, the x _i- axis of the image coordinate system extending to the right along the horizontal direction of the image, and the y _i- axis of the image coordinate system extending downward along the vertical direction of the image. The unit of the image coordinate system is pixel.

Furthermore, the Plücker coordinate representation specifically includes:

The equation of the straight line in the image space coordinate system can be described as:

ax _i +by _i +c = 0

Therefore, a straight line can be represented by a three-dimensional vector:

l＝[a,b,c] ^T

In the object space world coordinate system, the three-dimensional coordinates of two points A and B are Their homogeneous coordinates are: The line passing through these two points can be represented by a 4×4 antisymmetric homogeneous matrix L, which is called the Plücker matrix:

L＝AB ^T -BA ^T

Alternatively, the line L can be expressed by its direction vector It is expressed by moment m, which is called Plücker coordinates and is recorded as:

in, is the direction vector of the line, and the moment m is the normal vector of the plane determined by the line and the origin, that is,

From this, the relationship between the Plücker matrix and the Plücker coordinates is:

Under the action of the camera mapping T, the line L defined by the Plücker matrix represents the image l of the corresponding line in the image coordinate system:

Among them, K is the camera intrinsic parameter matrix:

is the attitude transfer matrix from the world coordinate system to the camera coordinate system, is the position vector of the origin of the camera coordinate system in the world coordinate system; s ² is just the common coefficient of the parameters in the straight line, so [l] _× can be simplified to:

Furthermore, the equations for calculating the vanishing point and the vanishing line at infinity in step (3) include:

During the landing process of the UAV, the visual landing system extracts the runway line features, where the left and right side lines and the center line of the runway are a set of parallel lines that can be used to calculate the vanishing point coordinates and vanishing line equations at infinity. Assume that the three equidistant parallel lines on the runway in the object space are L _0w , L _1w , and L _2w , and their images in the image plane are l _0i , l _1i , and l _2i , then the vanishing point coordinates in the image space can be solved by the following relationship:

The equation of the vanishing line is:

l _∞i = [(l _0i ×l _2i ) ^T (l _1i ×l _2i )]l _1i +2[(l _0i ×l _1i ) ^T (l _2i ×l _1i )]l _2i

Assume that the four-dimensional homogeneous coordinates of a point A in object space are Then it passes through point A and the direction is It can be expressed as:

When the parameter λ changes from 0 to ∞, point A changes from a finite point to an infinite point, and the coordinates of this point in the world coordinate system are:

According to the image conjugate equation, the relationship between the vanishing point and the UAV attitude transfer matrix is:

Further rearranging the equation yields:

Assume that a point on the image space blanking line _l∞i is x, and its reverse projection in the object space is a line with the direction A straight line; Since point x is on the straight line, we can get:

x ^T l _∞i ＝0

use Orthogonal to the plane's normal vector _nπ, we get:

Using the direction in object space The straight line projection of is the point x in the image space, and we get:

Transpose the above formula to get:

Combined with the previous text, we can get the image space vanishing line equation:

The attitude transfer matrix between the real-time world coordinate system and the camera coordinate system of the drone is solved by the simultaneous equations. And solve the attitude and lateral and vertical positions, including:

Combining the above formulas, we can get the following equations:

In the formula, is the attitude transfer matrix:

In the airport runway scenario, is the unit direction vector of the centerline of the runway in object space, _nπ is the unit normal vector of the track plane in object space, _nπ = [0,1,0] ^T , and similarly we can get

Let K ^-1 p _∞ =[g ₁ ,g ₂ ,g ₃ ] ^T , K ^T l _∞ =[h ₁ ,h ₂ ,h ₃ ] ^T ,(K ^-1 p _∞ )×(K ^T l _∞ )=[e ₁ ,e ₂ ,e ₃ ] ^T , and the attitude transfer matrix C _w c is an antisymmetric matrix, and the sum of the squares of each row and column element is 1. The above formula can be solved:

The three attitude angles are obtained from this:

Calculate relative position using line equation:

in:

Substitute the runway edge equation into the above formula:

Similarly, we can determine from the straight line _L2 :

Solve α ₀ and α ₂ from the above two equations:

Finally, the vertical and lateral positions _ty and _tx of the UAV in the world coordinate system are solved:

Furthermore, in step (4), the continuous state equation of the Kalman filter model of the inertial/visual integrated navigation system is as follows:

Where F(t) is the state transfer matrix of the continuous state equation at time t, is the system random noise vector at time t;

The filtering state quantities are north sky east velocity error, latitude error, altitude error, longitude error, north sky east misalignment angle error, gyro drift of the carrier system in XYZ direction, and accelerometer zero position of the carrier system in XYZ direction;

System state transition matrix

in:

The observation equation is defined as follows:

To observe the noise array;

The observation quantity of the integrated navigation system is the difference between the vertical lateral position output by the inertial navigation of the airport coordinate system and the navigation result of the visual landing system:

H(t)＝[0 _3×3 M _3×3 0 _3×9 ]

in,

The present invention proposes an infinite element and its Plücker coordinate representation method, which solves the problem that the coordinates of the intersection of parallel lines in the object space on the image plane cannot be solved. Secondly, the equation of the vanishing line at infinity is solved by a set of equidistant parallel lines on the left and right side lines and the center line of the runway, which solves the problem that the vanishing line cannot be identified by the feature extraction method. Finally, the UAV posture is solved by the vanishing line equation and integrated with the inertial, which solves the problem of the cumulative divergence of inertial navigation errors over time and the large noise in the visual navigation solution results.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 Schematic diagram of coordinate system;

Fig. 2 is a schematic diagram of a group of parallel lines intersecting at a vanishing point;

Fig. 3 Schematic diagram showing the meaning of the Plücker coordinates of a straight line.

DETAILED DESCRIPTION

The present invention will be further described in detail below in conjunction with the accompanying drawings.

Aiming at the autonomous landing navigation problem of UAVs under satellite denial conditions, the present invention conducts research on inertial/visual landing navigation methods based on line features. First, an infinite element and its Plücker coordinate representation method are proposed, which solves the problem that the coordinates of the intersection of parallel lines in the object space on the image plane cannot be solved. Secondly, the equation of the vanishing line at infinity is solved by a group of equidistant parallel lines on the left and right side lines and the center line of the runway, which solves the problem that the invisible line cannot be identified by the feature extraction method. Finally, the UAV posture is solved by the invisible line equation and integrated with the inertial, which solves the problem of the cumulative divergence of inertial navigation errors over time and the large noise in the visual navigation solution results.

1. Elements at Infinity and Plücker Notation

(1) Coordinate system definition

As shown in Figure 1, the airport coordinate system, visual coordinate system, world coordinate system, camera coordinate system, and image coordinate system are established.

Among them, the airport coordinate system (a system) takes the intersection of the start line of the runway landing end and the center line of the runway as the origin _oa ; the axis is along the runway center line, and the forward direction is positive; the _ya axis is perpendicular to the runway plane, and the upward direction is positive; the z _a axis coincides with the start line of the runway, and the right direction is positive; _oa x _a y _a z _a constitutes a right-handed coordinate system; the coordinates of a point in the airport coordinate system are expressed by (x _a , y _a , z _a ).

Visual coordinate system (V system): landing visual navigation system coordinate system, referred to as visual coordinate system; the image principal point of the optical system is taken as the origin O _V ; the x _V axis is parallel to the optical axis, and the forward direction is positive; the y _V axis is parallel to the horizontal axis of the imaging plane coordinate system, and the upward direction is positive; the z _V axis, the x _V axis and the y _V axis form a right-handed coordinate system, and the right direction is positive.

World coordinate system (W system): The intersection of the starting line of the aiming point at the landing end of the runway and the center line of the runway is the origin o _w ; the x _w axis coincides with the starting line of the runway, and the right direction is positive; the y _w axis is perpendicular to the runway plane, and the downward direction is positive; the z _w axis is along the center line of the runway, and the forward direction is positive; o _w x _w y _w z _w constitute a right-handed coordinate system; the coordinates of a point in the world coordinate system are expressed by (x _w , y _w , z _w ).

Camera coordinate system (C system): The image principal point of the optical system is taken as the origin _oc ; when observing the optical system directly, the _xc axis is parallel to the horizontal axis of the imaging plane coordinate system, and the left direction is positive; the _yc axis is parallel to the vertical axis of the imaging plane coordinate system, and the downward direction is positive; the _zc axis points to the observer and forms a right-handed coordinate system with the _xc axis and the _yc axis.

Image coordinate system (i system): The image coordinate system is established in the plane where the camera's photosensitive surface is located. It is a two-dimensional plane coordinate system with the upper left corner of the image as the origin. The x _i axis of the image coordinate system is to the right along the horizontal direction of the image, and the y _i axis of the image coordinate system is to the bottom along the vertical direction of the image. The unit of the image coordinate system is pixel.

(2) Elements at Infinity

In object space, two parallel lines never intersect. Based on Euclidean space, a projective space is constructed by introducing infinite elements. A set of parallel lines in a plane intersect at a unique point at infinity, which is called the vanishing point. As shown in Figure 2, the position of this point on the image plane is only related to the camera's posture and has nothing to do with the camera's position.

The vanishing point represents the direction of the corresponding parallel line. Unlike the infinity points of non-parallel lines, all the infinity points on the plane form a straight line, the vanishing line. The vanishing line is the only intersection line of a set of parallel planes in space at infinity.

(3) Plücker notation

The equation of the straight line in the image space coordinate system can be described as:

ax _i +by _i +c = 0

Therefore, a straight line can be represented by a three-dimensional vector:

l＝[a,b,c] ^T

In the object space world coordinate system, the three-dimensional coordinates of two points A and B are (3×1 matrix), then their homogeneous coordinates are: The straight line passing through these two points can be represented by a 4×4 antisymmetric homogeneous matrix L, which is called the Plücker matrix.

L＝AB ^T -BA ^T

Alternatively, the line L can be expressed by its direction vector It is expressed by moment m, which is called Plücker coordinates and is recorded as:

in, is the direction vector of the line, and the moment m (which can represent the area of ΔABC or the distance from O to the line L) is the normal vector of the plane determined by the line and the origin, that is, (As shown in Figure 3):

From this, the relationship between the Plücker matrix and the Plücker coordinates is:

Under the action of the camera mapping T, the line L defined by the Plücker matrix represents the image l of the corresponding line in the image coordinate system:

Among them, K is the camera intrinsic parameter matrix:

is the attitude transfer matrix from the world coordinate system to the camera coordinate system, is the position vector of the origin of the camera coordinate system in the world coordinate system. ^{s 2} is just the common coefficient of the parameters in the straight line, so [l] _× can be simplified to:

2. Visual measurement pose calculation

(1) Vanishing point and vanishing line imaging equations

During the landing process of the drone, the visual landing system extracts the runway line features, where the left and right side lines and the center line of the runway are a set of parallel lines that can be used to calculate the vanishing point coordinates and vanishing line equations at infinity. Assume that the three equidistant parallel lines on the runway in the object space are L _0w , L _1w , and L _2w , and their images in the image plane are l _0i , l _1i , and l _2i , then the vanishing point coordinates in the image space can be solved by the following relationship:

The equation of the vanishing line is:

l _∞i = [(l _0i ×l _2i ) ^T (l _1i ×l _2i )]l _1i +2[(l _0i ×l _1i ) ^T (l _2i ×l _1i )]l _2i

Assume that the four-dimensional homogeneous coordinates of a point A in object space are Then it passes through point A and the direction is

(three-dimensional unit column vector), which can be expressed as:

When the parameter λ changes from 0 to ∞, point A changes from a finite point to an infinite point, and the coordinates of this point in the world coordinate system are:

According to the image conjugate equation, the relationship between the vanishing point and the UAV attitude transfer matrix is:

Further rearranging the equation yields:

Assume that a point on the image space blanking line _l∞i is x, and its reverse projection in the object space is a line with the direction Since point x is on the straight line, we can get:

x ^T l _∞i ＝0

use Orthogonal to the plane's normal vector _nπ, we get:

Using the direction in object space The straight line projection of is the point x in the image space, and we get:

Transpose the above formula to get:

Combined with the previous text, we can get the image space vanishing line equation:

(2) UAV posture calculation

Combining the formulas in the previous article, we can get the following system of equations:

In the formula, is the attitude transfer matrix:

In the airport runway scenario, is the unit direction vector of the centerline of the runway in object space, _nπ is the unit normal vector of the track plane in object space, _nπ = [0,1,0] ^T , and similarly we can get

Let K ^-1 p _∞ =[g ₁ ,g ₂ ,g ₃ ] ^T , K ^T l _∞ =[h ₁ ,h ₂ ,h ₃ ] ^T ,(K ^-1 p _∞ )×(K ^T l _∞ )=[e ₁ ,e ₂ ,e ₃ ] ^T , and the attitude transfer matrix C _w c is an antisymmetric matrix, and the sum of the squares of each row and column element is 1. The above formula can be solved:

The three attitude angles are obtained from this:

Calculate relative position using line equation:

in:

Substitute the runway edge equation into the above formula:

Similarly, we can determine from the straight line _L2 :

Solve α ₀ and α ₂ from the above two equations:

Finally, the vertical and lateral positions _ty and _tx of the UAV in the world coordinate system are solved:

3. Inertial/visual fusion method

The continuous state equation of the Kalman filter model of the inertial/visual integrated navigation system is as follows:

Where F(t) is the state transfer matrix of the continuous state equation at time t, is the system random noise vector at time t.

The filtering state quantities are north sky east velocity error (unit: m/s), latitude error (unit: rad), altitude error (unit: m), longitude error (unit: rad), north sky east misalignment angle error (unit: rad), gyro drift of the carrier system in the XYZ direction (unit: rad/s), and accelerometer zero position of the carrier system in the XYZ direction (unit: m/ ^s2 ).

System state transition matrix

in:

The observation equation is defined as follows:

is the observed noise array.

The observation quantity of the integrated navigation system is the difference between the vertical lateral position output by the inertial navigation of the airport coordinate system and the navigation result of the visual landing system:

H(t)＝[0 _3×3 M _3×3 0 _3×9 ]

in,

In summary, an inertial/visual navigation method using three equidistant parallel lines, the left and right side lines and the center line of the runway, is proposed during the landing phase of the UAV.

The present invention specifically realizes autonomous landing of the drone through the following four processes:

(1) Airport runway data collection:

The forward-looking infrared vision navigation system installed on the nose of the UAV is used to collect images of the airport runway, and the real-time feature extraction of the runway sideline is performed to obtain the straight line equations of the sideline and centerline.

(2) Plücker coordinate representation:

The Plücker coordinates of the edge equation are calculated according to the formula derived above by using the pre-bound airport runway width and the camera intrinsic parameter matrix.

(3) Visual measurement pose solution:

After obtaining the Plücker coordinates of the runway edge, the equations of the vanishing point and the vanishing line at infinity are calculated by two equidistant parallel lines, and the attitude transfer matrix between the real-time world coordinate system and the camera coordinate system of the drone is solved by the simultaneous equations. And solve the attitude and lateral and vertical positions.

(4) Combined navigation based on inertial/visual fusion:

The position information calculated by the visual landing system is used as the observation quantity, and the Kalman filter is constructed to fuse it with the navigation information output by the inertial navigation to realize continuous and autonomous navigation and positioning functions.

Claims (5)

1. A UAV inertial/visual landing navigation method based on line features, characterized in that it includes the following steps: (1) Airport runway data collection Establish the airport coordinate system, visual coordinate system, world coordinate system, camera coordinate system, and image coordinate system; collect images of the airport runway, and perform real-time feature extraction on the runway sidelines to obtain the linear equations of the sidelines and centerlines; (2) Plücker coordinate representation The Plücker coordinates of the edge equation are calculated by using the pre-bound airport runway width and the camera internal parameter matrix; (3) Visual measurement pose calculation After obtaining the Plücker coordinates of the runway edge, the equations of the vanishing point and the vanishing line at infinity are calculated by two equidistant parallel lines, and the attitude transfer matrix between the real-time world coordinate system and the camera coordinate system of the drone is solved by the simultaneous equations. And solve the attitude and lateral and vertical positions; (4) Combined navigation based on inertial/visual fusion The position information calculated by the visual landing system is used as the observation quantity, and the Kalman filter is constructed and fused with the navigation information output by the inertial navigation to realize the continuous autonomous navigation and positioning function; the observation quantity is the difference between the vertical lateral position output by the inertial navigation of the airport coordinate system and the navigation result of the visual landing system. 2. The UAV inertial/visual landing navigation method based on line features according to claim 1 is characterized in that the airport coordinate system, visual coordinate system, world coordinate system, camera coordinate system and image coordinate system are established in step (1), including: The airport coordinate system is denoted as the a system; the intersection of the runway landing end start line and the runway center line is the origin _oa ; the axle is along the runway center line, and the forward direction is positive; the _{ya axis is perpendicular to the runway plane, and the upward direction is positive; the z a axis} coincides with the runway start line, and the right direction is positive; _oa x _a y _a z _a constitutes a right-handed coordinate system; the coordinates of a point in the airport coordinate system are expressed as (x _a , y _a , z _a ); Visual coordinate system, denoted as v system; the image principal point of the optical system is taken as the origin o _v ; the x _v axis is parallel to the optical axis, and the forward direction is positive; the y _v axis is parallel to the horizontal axis of the imaging plane coordinate system, and the upward direction is positive; the z _v axis, the x _v axis and the y _v axis form a right-handed coordinate system, and the right direction is positive; The world coordinate system is recorded as the w system; the intersection of the runway landing end aiming point starting line and the runway centerline is taken as the origin o _w ; the x _w axis coincides with the runway starting line, and the right direction is positive; the y _w axis is perpendicular to the runway plane, and the downward direction is positive; the z _w axis is along the runway centerline, and the forward direction is positive; o _w x _w y _w z _w constitutes a right-handed coordinate system; the coordinates of a point in the world coordinate system are expressed as (x _w ,y _w ,z _w ); The camera coordinate system is denoted as the c system; the image principal point of the optical system is taken as the origin o _c ; when observing the optical system directly, the x _c axis is parallel to the horizontal axis of the imaging plane coordinate system, and the left direction is positive; the y _c axis is parallel to the vertical axis of the imaging plane coordinate system, and the downward direction is positive; the z _c axis points to the observer and forms a right-handed coordinate system with the x _c axis and the y _c axis; The image coordinate system is denoted as the i-system; the image coordinate system is established in the plane where the camera's photosensitive surface is located, with the upper left corner of the image as the origin, the x _i- axis of the image coordinate system extending to the right along the horizontal direction of the image, and the y _i- axis of the image coordinate system extending downward along the vertical direction of the image. The unit of the image coordinate system is pixel. 3. The UAV inertial/visual landing navigation method based on line features according to claim 2, characterized in that the Plücker coordinate representation specifically includes: The equation of the straight line in the image space coordinate system can be described as: ax _i +by _i +c = 0 Therefore, a straight line can be represented by a three-dimensional vector: l＝[a,b,c] ^T In the object space world coordinate system, the three-dimensional coordinates of two points A and B are Their homogeneous coordinates are: The line passing through these two points can be represented by a 4×4 antisymmetric homogeneous matrix L, which is called the Plücker matrix: L＝AB ^T -BA ^T Alternatively, the line L can be expressed by its direction vector It is expressed by moment m, which is called Plücker coordinates and is recorded as: in, is the direction vector of the line, and the moment m is the normal vector of the plane determined by the line and the origin, that is, From this, the relationship between the Plücker matrix and the Plücker coordinates is: Under the action of the camera mapping T, the line L defined by the Plücker matrix represents the image l of the corresponding line in the image coordinate system: Among them, K is the camera intrinsic parameter matrix: is the attitude transfer matrix from the world coordinate system to the camera coordinate system, is the position vector of the origin of the camera coordinate system in the world coordinate system; s ² is just the common coefficient of the parameters in the straight line, so [l] _× can be simplified to: 4. The UAV inertial/visual landing navigation method based on line features according to claim 3 is characterized in that the equations for calculating the vanishing point and the vanishing line at infinity in step (3) include: During the landing process of the UAV, the visual landing system extracts the runway line features, where the left and right side lines and the center line of the runway are a set of parallel lines that can be used to calculate the vanishing point coordinates and vanishing line equations at infinity. Assume that the three equidistant parallel lines on the runway in the object space are L _0w , L _1w , and L _2w , and their images in the image plane are l _0i , l _1i , and l _2i , then the vanishing point coordinates in the image space can be solved by the following relationship: The equation of the vanishing line is: l _∞i = [(l _0i ×l _2i ) ^T (l _1i ×l _2i )]l _1i +2[(l _0i ×l _1i ) ^T (l _2i ×l _1i )]l _2i Assume that the four-dimensional homogeneous coordinates of a point A in object space are Then it passes through point A and the direction is It can be expressed as: When the parameter λ changes from 0 to ∞, point A changes from a finite point to an infinite point, and the coordinates of this point in the world coordinate system are: According to the image conjugate equation, the relationship between the vanishing point and the UAV attitude transfer matrix is: Further rearranging the equation yields: Assume that a point on the image space blanking line _l∞i is x, and its reverse projection in the object space is a line with the direction A straight line; Since point x is on the straight line, we can get: x ^T l _∞i ＝0 use Orthogonal to the plane's normal vector _nπ, we get: Using the direction in object space The straight line projection of is the point x in the image space, and we get: Transpose the above formula to get: Combined with the previous text, we can get the image space vanishing line equation: The attitude transfer matrix between the real-time world coordinate system and the camera coordinate system of the drone is solved by the simultaneous equations. And solve the attitude and lateral and vertical positions, including: Combining the above formulas, we can get the following equations: In the formula, is the attitude transfer matrix: In the airport runway scenario, is the unit direction vector of the centerline of the runway in object space, _nπ is the unit normal vector of the track plane in object space, _nπ = [0,1,0] ^T , and similarly we can get Let K ^-1 p _∞ = [g ₁ , g ₂ , g ₃ ] ^T , K ^T l _∞ = [h ₁ , h ₂ , h ₃ ] ^T , (K ^-1 p _∞ ) × (K ^T l _∞ ) = [e ₁ , e ₂ , e ₃ ] ^T , and the attitude transfer matrix It is an antisymmetric matrix, and the sum of the squares of each row and column element is 1. The above formula can be solved: The three attitude angles are obtained from this: Calculate relative position using line equation: in: Substitute the runway edge equation into the above formula: Similarly, we can determine from the straight line _L2 : Solve α ₀ and α ₂ from the above two equations: Finally, the vertical and lateral positions _ty and _tx of the UAV in the world coordinate system are solved: 5. The method for unmanned aerial vehicle inertial/visual landing navigation based on line features according to claim 4 is characterized in that, in step (4), the continuous state equation of the Kalman filter model of the inertial/visual integrated navigation system is as follows: Where F(t) is the state transfer matrix of the continuous state equation at time t, is the system random noise vector at time t; The filtering state quantities are north sky east velocity error, latitude error, altitude error, longitude error, north sky east misalignment angle error, gyro drift of the carrier system in XYZ direction, and accelerometer zero position of the carrier system in XYZ direction; System state transition matrix in: The observation equation is defined as follows: To observe the noise array; The observation quantity of the integrated navigation system is the difference between the vertical lateral position output by the inertial navigation of the airport coordinate system and the navigation result of the visual landing system: H(t)＝[0 _3×3 M _3×3 0 _3×9 ] in, .