CN115683102A - Unmanned agricultural machinery navigation method, equipment, device and storage medium - Google Patents

Abstract

The invention discloses a navigation method, device, device and storage medium for unmanned agricultural machinery. The method includes: determining the geodetic coordinates of the characteristic corner points of the graphical signboards arranged on the driving path of the unmanned agricultural machinery or on both sides of the driving path; The geodetic coordinates of the unmanned agricultural machine are obtained by positioning during the driving process of the unmanned agricultural machine, the surrounding environment image is collected, and the edge contour of the pattern in the surrounding environment image is extracted using the image edge detection technology, thereby identifying the pattern in the surrounding environment image The apex, using the apex as a feature corner point, calculates the pixel coordinates of the feature corner point; uses the geodetic coordinates and pixel coordinates of the feature corner point to correct and calibrate the geodetic coordinates of the unmanned agricultural machine. This method can use computer vision technology to correct the position coordinates of unmanned agricultural machinery without relying on vehicle-mounted RTK equipment and technology, so as to realize high-precision positioning of centimeter-level unmanned agricultural machinery, reduce equipment costs, and improve positioning and navigation accuracy.

Description

Unmanned agricultural machinery navigation method, equipment, device and storage medium

technical field

The invention relates to the fields of computer vision and assisted automatic driving. More specifically, it relates to a navigation method, equipment, device and storage medium based on unmanned agricultural machinery.

Background technique

my country is a large agricultural country with a vast area of cultivated land, accounting for 7% of the world's cultivated land. However, compared with developed countries such as Europe and the United States, our current level of agricultural mechanization is still relatively low, and the level of industrial development needs to be improved. At the same time, with the deepening of urbanization, more and more young rural laborers leave their rural areas to work in cities, resulting in a continuous decrease of young rural laborers, which has brought a great impact on agricultural operations and the rural economy. Therefore, unmanned agricultural machinery has become one of the important landing scenarios of autonomous driving (unmanned driving) technology in recent years. With the support of automatic driving technology, agricultural machinery can realize diversified operations such as automatic plowing, leveling, sowing, and harvesting, which can greatly make up for the lack of young labor and professional drivers, and play an important role in improving the efficiency and automation level of farming operations. significance.

Due to the need for precise operations, unmanned agricultural machinery has high requirements for navigation and positioning accuracy. Generally, it is necessary to use real-time dynamic (RTK, Real-Time Kinematic) carrier phase difference measurement equipment on unmanned agricultural machinery to obtain and Maintain centimeter-level high-precision positioning capabilities, thereby providing the basis for operations such as driving, control, and path planning of agricultural machinery. However, the widespread application of RTK devices in farming is still limited. First of all, the cost of using RTK is still relatively high. When using third-party commercial services, users need to pay a lot of equipment fees and service fees. When the number of agricultural machinery and equipment is large, this cost will be more significant. Secondly, in many remote areas, RTK service coverage is still not available. At this time, if users set up RTK reference stations and communication facilities by themselves, they need to build and maintain them by themselves, which will increase the cost of manpower and equipment. Therefore, how to obtain an accurate, reliable, and cost-controllable high-precision position reference has become one of the outstanding pain points in the large-scale application of unmanned agricultural machinery.

Contents of the invention

In order to solve the above-mentioned technical problems or at least partly solve the above-mentioned technical problems, the present disclosure provides an unmanned agricultural machinery navigation method, equipment and storage medium, which are used to realize the navigation of unmanned agricultural machinery without relying on vehicle-mounted RTK equipment and technology. Accurate and reliable navigation and positioning.

An embodiment of the present invention provides a method for navigating an unmanned agricultural machine, the method comprising: determining the geodetic coordinates of the characteristic corner points of the graphical signboards arranged on the driving path of the unmanned agricultural machine or on both sides of the driving path; The geodetic coordinates of the unmanned agricultural machine are obtained by positioning in the center, the surrounding environment image is collected, and the edge contour of the pattern in the surrounding environment image is extracted by using the image edge detection technology, thereby identifying the vertices of the pattern in the surrounding environment image, and the The apex is used as a characteristic corner point, and the pixel coordinates of the characteristic corner point are calculated; the earth coordinates of the unmanned agricultural machine are corrected and calibrated by using the earth coordinates and pixel coordinates of the characteristic corner point.

Preferably, the method for determining the characteristic angles of the graphic signboards arranged on the driving path of the unmanned agricultural machinery or on both sides of the driving path according to claim 1, the geodetic coordinates of the points comprises: utilizing the RTK high-precision receiver to set The static measurement collects the coordinate positions of the graphic signage, and takes an average value of the coordinate positions of the graphic signpost collected in the period of time, and calculates the geodetic coordinates of the characteristic corner points of the graphic signboard.

Preferably, using image edge detection technology, extracting the edge profile of the pattern in the surrounding environment image includes calculating the gradient and angle of the pattern in the surrounding environment image; performing non-maximum suppression on the gradient; The edges are connected until the complete edge contour of the pattern is extracted.

其中，x为所述周边环境图像中图案的像素点的横坐标，y为所述周边环境图像中图案的像素点的纵坐标，f为所述周边环境图像中图案的灰度值，

为所述周边环境图像中图案的所有像素点计算形成的角度矩阵。Preferably, calculating the gradient and angle of the pattern in the surrounding environment image includes: calculating the gradient and angle of the pattern in the surrounding environment image according to the following formula,

Wherein, x is the abscissa of the pixel of the pattern in the surrounding environment image, y is the ordinate of the pixel of the pattern in the surrounding environment image, f is the gray value of the pattern in the surrounding environment image,

The formed angle matrix is calculated for all pixels of the pattern in the surrounding environment image.

Preferably, performing non-maximum value suppression on the gradient includes: judging whether the gray value of the currently detected point C in the gradient is the largest in the 8-connected neighborhood, and if it is the largest, continue to check the gradient in the gradient Whether the gray value of the first intersection point dTmp1 and the second corner point dTmp2 of the direction is greater than C, if C is greater than the gray value of the first intersection point dTmp1 and the second corner point dTmp2, then C is considered to be the maximum value and the value of C Set it to 1, otherwise it is considered that C is a non-maximum value and the value of C is set to 0, and all points C in the gradient are traversed to find the local maximum value of the pixel point in the gradient, and the non-maximum value of the gradient is completed value suppression.

Preferably, using double thresholds to connect the edges of the pattern until the complete edge profile of the pattern is extracted includes selecting two thresholds, the two thresholds include a low threshold and a high threshold, and points less than the low threshold are considered false Edges are concatenated to 0, and points greater than the high threshold are considered strong edges and concatenated to 1; according to the high threshold points in the image, they are first connected into contours, and when the breakpoint of the contour is reached, the algorithm will be at the breakpoint 8 in the field to find a point that satisfies the low threshold, and then collect new edges based on this point until the entire image is closed.

其中，p′_j为所述特征角点的的大地坐标，p_j为待优化的无人农机的修正坐标，z′_ij为无人农机在位姿T_i处观察所述特征角点p′_j所产生的像素测量数据，e_ij为误差函数，h(T_i，pj)为世界坐标到像素坐标的投影函数。Preferably, using the geodetic coordinates and pixel coordinates of the feature corner points, correcting and calibrating the geodetic coordinates of the unmanned agricultural machine includes: calculating the corrected coordinates of the unmanned agricultural machine according to the following formula,

Among them, p' _j is the geodetic coordinate of the characteristic corner point, p _j is the correction coordinate of the unmanned agricultural machine to be optimized, and z' _ij is the characteristic corner point p' observed by the unmanned agricultural machine at the pose T _i The pixel measurement data generated by _j , e _ij is an error function, and h(T _i , pj) is a projection function from world coordinates to pixel coordinates.

On the other hand, an embodiment of the present invention provides a navigation device for unmanned agricultural machinery, wherein the device includes: a device for determining the geodetic coordinates of characteristic corner points, which is used to determine the location on the driving path of the unmanned agricultural machinery or on both sides of the driving path. The geodetic coordinates of the characteristic corner points of the graphic signage; the pixel coordinate calculation device of the characteristic corner points obtains the geodetic coordinates of the unmanned agricultural machinery during the driving process, collects the surrounding environment images, and uses the image edge detection technology to extract the said The edge contour of the pattern in the surrounding environment image, thereby identifying the vertex of the pattern in the surrounding environment image, using the vertex as a feature corner point, and calculating the pixel coordinates of the feature corner point; the coordinate calibration device is used to use The geodetic coordinates and pixel coordinates of the feature corner points correct and calibrate the geodetic coordinates of the unmanned agricultural machine.

In another aspect, the embodiment of the present invention also provides a navigation device for unmanned agricultural machinery, the device includes: a processor, a memory, including program instructions executable by the processor, when the program instructions are executed by the processor, Make the device for measuring the position coordinates of the indoor wireless signal transmitting anchor point perform the following operations: determine the geodetic coordinates of the characteristic corner points of the graphic signboards arranged on the driving path of the unmanned agricultural machine or on both sides of the driving path; The geodetic coordinates of the unmanned agricultural machine are obtained by positioning during the driving of the agricultural machine, the surrounding environment image is collected, and the edge contour of the pattern in the surrounding environment image is extracted using the image edge detection technology, thereby identifying the vertices of the pattern in the surrounding environment image, The vertex is used as a feature corner point, and the pixel coordinates of the feature corner point are calculated; using the geodetic coordinates and pixel coordinates of the feature corner point, the geodetic coordinates of the unmanned agricultural machine are corrected and calibrated.

In yet another aspect, an embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored, wherein the computer program implements any one of claims 1 to 7 when executed by a processor. Methods.

Compared with the prior art, the technical solutions provided by the embodiments of the present disclosure have the following advantages: the present disclosure sets graphic signs with characteristic corners on both sides of the driving path of the unmanned agricultural machine, and uses image edge detection technology during driving , identify the characteristic corners of the graphic sign and calculate the pixel coordinates of the characteristic corners, and then use the geodetic coordinates and pixel coordinates of the characteristic corners to correct and calibrate the geodetic coordinates of the unmanned agricultural machine, so that it can be In the case of relying on vehicle-mounted RTK equipment and technology, computer vision technology is used to correct the position coordinates of unmanned agricultural machinery to achieve high-precision positioning of centimeter-level unmanned agricultural machinery, which meets the high-precision operation requirements of unmanned agricultural machinery, greatly reduces equipment costs, and improves positioning and navigation accuracy.

Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description.

Description of drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description serve to explain the principles of the disclosure.

In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, In other words, other drawings can also be obtained from these drawings without paying creative labor.

FIG. 1 is a flow chart of a navigation method for unmanned agricultural machinery described in an embodiment of the present disclosure;

Fig. 2 is an exemplary graphic sign according to an embodiment of the present disclosure;

Fig. 3 is a schematic diagram of setting up a graphic sign according to an embodiment of the present disclosure;

FIG. 4 is a schematic diagram of non-maximum suppression of gradients according to an embodiment of the present disclosure;

5 is a structural block diagram of an unmanned agricultural machinery navigation device according to an embodiment of the present disclosure;

Fig. 6 is a schematic diagram of a navigation device for an unmanned agricultural machine according to an embodiment of the present disclosure.

Detailed ways

In order to enable those skilled in the art to better understand the technical solutions of the present disclosure, the present disclosure will be described in detail below in conjunction with the accompanying drawings and specific embodiments. Embodiments of the present disclosure will be described in further detail below in conjunction with the accompanying drawings and specific embodiments, but are not intended to limit the present disclosure.

With the rapid construction of my country's independent Beidou satellite navigation system, the Beidou-3 satellite navigation system has begun to provide open and free sub-meter-level satellite-based augmentation (Satellite-Based Augmentation System, SBAS) services and decimeter-level precision to the industry and the public. Single point positioning (Precise Point Positioning, PPP) service. In addition, unmanned agricultural machinery is generally equipped with or already equipped with inertial navigation (Inertial Navigation System, INS), camera (Camera) sensors and other equipment. Therefore, the inventor proposes that by integrating the Beidou SBAS/PPP service and its navigation methods with inertial and visual (Visual), it is possible for unmanned agricultural machinery to obtain continuous, stable and reliable navigation and positioning results with decimeter-level absolute precision.

Based on this, as shown in Figure 1, the present invention provides a kind of unmanned agricultural machine navigation method, and this method comprises:

S101, determining the geodetic coordinates of the characteristic corner points of the graphical signage set on the driving path of the unmanned agricultural machine or on both sides of the driving path;

Among them, on the driving path of the unmanned agricultural machinery or on both sides of the driving path, set up graphic signs with characteristic corner points that can be clearly identified by computer vision.

In some embodiments, the graphic indicator may be in the form of a square, a rectangle, a triangle, or a two-dimensional code, as shown in FIG. 2 .

When the unmanned agricultural machine works in the farmland, it often adopts the method of A/B point operation, and moves from point A to point B according to the preset driving path, as shown in Figure 3. Therefore, in some embodiments, the graphic signage can be arranged on the driving path of the unmanned agricultural machine, such as the end of the ridge; it can also be arranged on both sides of the driving path of the unmanned agricultural machine, such as both sides of the ridge.

In a preferred embodiment, in order to accurately obtain the geodetic coordinates of the graphic signboard and correct the positioning results of the subsequent unmanned agricultural machinery, it is possible to use, for example, a high-precision satellite positioning receiver to collect static measurements within a period of time in advance. For the coordinate position of the graphic indicator, take the average value of the coordinate positions of the graphic indicator collected in the period of time, and calculate the geodetic coordinates of the characteristic corner points of the graphic indicator.

Further preferably, when collecting the coordinate position of the characteristic corner point, it is necessary to make the phase center of the antenna of the satellite positioning receiver coincide with the characteristic corner point of the graphic signboard as much as possible, so as to ensure accurate measurement of the ground of the characteristic corner point of the image signboard coordinate. For example, place the satellite positioning receiver antenna on the vertices of a square or rectangular sign. If the phase center of the satellite positioning receiver antenna is difficult to accurately coincide with the characteristic corner point, it can also be calibrated and corrected manually by the surveyor.

S102. Locate and obtain the geodetic coordinates of the unmanned agricultural machine during driving, collect the surrounding environment image, and use the image edge detection technology to extract the edge contour of the pattern in the surrounding environment image, thereby identifying the surrounding environment image The vertex of the pattern is used as a feature corner point, and the pixel coordinates of the feature corner point are calculated;

Wherein, in some embodiments, the two-dimensional pixel coordinates of the feature corners can be obtained by processing in the following manner: the surrounding environment image is collected by the camera sensor mounted on the unmanned agricultural machine, and for each frame of image, image edge detection technology is used , extract the edge profile of the image or pattern, thereby identify the vertices of the image or pattern, and use them as feature corners. The edge detection operator can choose Canny operator, Roberts operator, Sobel operator, Marr-Hildreth operator, etc.

In order to find an optimal edge, for example, identify as many actual edges in the image as possible; the identified edge should be as close as possible to the edge in the actual image; the edge in the image can only be identified once, and there may be The image noise should not be identified as an edge. In a preferred embodiment, using image edge detection technology, extracting the edge profile of the pattern in the surrounding environment image may include the following steps:

S201. Calculate gradients and angles of patterns in the surrounding environment image.

S202, performing non-maximum suppression on the gradient;

S203, using double thresholds to connect the edges of the pattern until a complete edge profile of the pattern is extracted.

In step S201, the gradient is a very important concept in artificial intelligence, which spreads in the fields of machine learning and deep learning. The first-order differential of a one-dimensional function is defined as formula (1):

Wherein, f(x) is a microfunction about the unknown x, x is the unknown, and ε is a tiny variable about the unknown.

The filtering of the image is generally based on the grayscale image, and the image is two-dimensional at this time. Therefore, two-dimensional function differential processing is required, that is, formula (2) and formula (3):

Where f(x, y) is a two-dimensional function about the unknowns x, y.

It can be seen from the above formula that the image gradient is the partial derivative of the current pixel point with respect to the X-axis and Y-axis, so the gradient can also be understood as the change speed of the pixel gray value in the field of image processing.

The modulus of the gradient represents the amount that f(x,y) increases per unit distance in the direction of its maximum rate of change, namely:

Among them, G is the modulus of the image gradient.

The angle calculation of the gradient is relatively simple, and its function provides a basis for the direction of non-maximum suppression. Calculated as follows:

为所有像素点计算形成的角度矩阵。Among them, x is the abscissa of the pixel in the image, y is the ordinate of the pixel in the image, f is the gray value of the image,

Calculate the resulting angle matrix for all pixels.

In step S202, non-maximum suppression is performed on the gradient.

The gradient obtained in step S201 has problems such as thick edges and weak edge interference. In this regard, the non-maximum value suppression process can be used to find the local maximum value of the pixel, and the gray value corresponding to the non-maximum value is set to 0, so that a large number of non-edge pixels can be eliminated.

As shown in Figure 4, C represents the point currently being detected, and g1-g4 are its 8-connected neighbor points. In the figure, the oblique line indicates the gradient direction of point C calculated in the previous step.

Judging whether the gray value of the currently detected point C in the gradient is the largest in the 8-connected neighborhood, if it is the largest, continue to check the gray values of the first intersection point dTmp1 and the second corner point dTmp2 of the gradient direction in the gradient Whether the value is greater than C, if C is greater than the gray value of the first intersection point dTmp1 and the second corner point dTmp2, then consider C to be a maximum value and set the value of C to 1, otherwise consider C to be a non-maximum value and set C The value of is set to 0, and all points C in the gradient are traversed, so as to find the local maximum of the pixel points in the gradient, and complete the non-maximum suppression of the gradient.

Among them, it should be noted that the intersection point of the gradient direction does not necessarily fall at the position of the 8 points in the 8 neighborhood, so the actual application of dTmp1 and dTmp2 is to use the grayscale formed by the bilinear interpolation of two adjacent points value.

In step S203, after the processing of steps S201 and S202, the edge quality of the pattern is already very high, but there are still many false edges. Therefore, in order to remove false edges, double thresholds can be used to connect the edges of the pattern. Specifically, two thresholds are selected, the two thresholds include a low threshold and a high threshold, points smaller than the low threshold are considered false edges and set to 0, and points greater than the high threshold are considered strong edges and set to 1 , pixels in between need further inspection.

According to the high-threshold points in the image, first connect them into contours. When the breakpoint of the contour is reached, it will find a point that meets the low threshold in the 8 fields of the breakpoint, and then collect new edges based on this point until the entire image closure.

S204, on the edge contour line of the pattern, identify and calculate the pixel coordinates of the feature corner points.

大于某一阈值Γ，即有：

时，则像素点j为所述图案的所述特征角点。For a certain pixel point j on the edge contour line, its pixel coordinates are marked as (uj, vj), and the pixel coordinates of its left and right adjacent pixel points are (uj-1, vj-1) and (uj+1, vj+1). like

greater than a certain threshold Γ, that is:

, then the pixel point j is the characteristic corner point of the pattern.

Wherein, the threshold Γ should be reasonably determined according to the geometric shape of the graphic sign. For example, when the image sign is a square or a rectangle, the threshold Γ should be set close to 90°; when the image sign is an equilateral triangle, the threshold Γ should be set close to 60°.

Further preferably, before step S201, Gaussian filtering may be used to perform denoising processing on the collected surrounding environment image. Gaussian filtering is a linear smoothing filter that can be used to eliminate Gaussian white noise and is widely used in noise reduction in image processing.

S103. Correct and calibrate the geodetic coordinates of the unmanned agricultural machine by using the geodetic coordinates and pixel coordinates of the feature corner points.

At this point, it is actually a process of establishing and solving a BA (Bundle Adjustment) model with constraints. For a classic Simultaneous Location and Mapping (SLAM) problem, we start from a point p in a world coordinate system, take into account the internal and external parameters and distortion of the camera mounted on the unmanned agricultural machine, and finally project To generate pixel coordinates, the following steps are required:

S301, convert the world coordinates to the camera coordinates, where the camera extrinsic parameters (R, t) will be used:

P'=Rp+t=[X', Y', Z'] ^T (6)

Among them, P' is the point p in the world coordinate system that falls on the physical imaging plane O'-x'-y' after being projected through the small hole O, and [X', Y', Z'] ^T is the point of P' Coordinates, R is the rotation matrix from world coordinates to camera coordinates, t is the translation vector from world coordinates to camera coordinates.

S302, projecting P′ onto a normalized plane to obtain normalized coordinates:

P _c =[u _c ,v _c ,1] ^T =[X'/Z',Y'/Z',1] ^T (7)

Wherein, P _c is the projected point of point P′ on the normalized pixel plane, and [u _c , v _c , 1] ^T is the coordinate of P _c .

S303, consider the distortion of the normalized coordinates, and obtain the original pixel coordinates [u′ _c , v′ _c ] ^T before de-distortion. Here, only the radial distortion is considered for the time being:

Among them, k ₁ , k ₂ , and _rc are distortion correction polynomial parameters.

S304, calculate pixel coordinates according to the internal reference model:

The above process can be abstractly recorded as formula (10):

z=h(x,y) (10)

基于最小二乘方法原理，可以列出此次观测的误差方程：The detailed parameterization process of BA processing is given above. Specifically, x here refers to the pose of the camera at this time, that is, the external parameters R and t, and its corresponding Lie group is denoted as T. The road sign feature point y is the three-dimensional point p here, and the observation data is the pixel coordinates

Based on the principle of least squares method, the error equation of this observation can be listed as follows:

e=z-h(T,p) (11)

Then, close measurements at other times are taken into account and a subscript is added to the error. Let z _ij be the data generated by the unmanned agricultural machine observing the landmark p _j at the pose T _i , then the overall cost function is

Solving the formula (12) is equivalent to adjusting the pose of the unmanned agricultural machine and the landmark features in the environment at the same time, which is the so-called BA.

On this basis, for the method of the present invention, record the identifiable characteristic corner point in the graphic sign as p′ _j , and z′ _ij is the characteristic corner point p′ _j observed by the unmanned agricultural machine at the pose T _i For the generated data, since the geodetic coordinates of the characteristic corner point p′ _j are known, it provides us with the conditions for correction and constraints, so we can add p′ _j and z′ on the basis of formula (6). The constraint equation of _ij , and rewrite the cost function as formula (13):

Formula (13) is not a linear function, and it can be solved by nonlinear optimization to finally obtain the high-precision navigation and positioning results of unmanned agricultural machinery based on the vision correction of the present invention.

Thus, the coordinates of the corrected unmanned agricultural machine are calculated according to formula (13),

Among them, p' _j is the geodetic coordinate of the characteristic corner point, p _j is the correction coordinate of the unmanned agricultural machine to be optimized, and z' _ij is the characteristic corner point p' observed by the unmanned agricultural machine at the pose T _i The pixel measurement data generated by _j , eij is an error function, and h(T _i , pj) is a projection function from world coordinates to pixel coordinates.

It is also worth pointing out that since the graphic signage will be continuously visible for a period of time during the driving of the unmanned agricultural machine, the effects of the above corrections and constraints are continuous during this period of time. If the graphic signs on the driving road of the unmanned agricultural machinery are appropriately increased, the method of the present invention can obtain better performance and effect.

Compared with the prior art, the technical solutions provided by the embodiments of the present disclosure have the following advantages: the present disclosure sets graphic signs with characteristic corners on both sides of the driving path of the unmanned agricultural machine, and uses image edge detection technology during driving , identify the characteristic corners of the graphic sign and calculate the pixel coordinates of the characteristic corners, and then use the geodetic coordinates and pixel coordinates of the characteristic corners to correct and calibrate the geodetic coordinates of the unmanned agricultural machine, so that it can be In the case of relying on vehicle-mounted RTK equipment and technology, computer vision technology is used to correct the position coordinates of unmanned agricultural machinery to achieve high-precision positioning of centimeter-level unmanned agricultural machinery, which meets the high-precision operation requirements of unmanned agricultural machinery, greatly reduces equipment costs, and improves positioning and navigation accuracy.

The embodiment of the present invention also provides an unmanned agricultural machine navigation device, wherein the device includes:

The feature corner geodetic coordinate determining device is used to determine the geodetic coordinates of the feature corner points of the graphic signage set on the driving path of the unmanned agricultural machinery or on both sides of the driving path;

The feature corner point pixel coordinate calculation device locates the unmanned agricultural machine to obtain the geodetic coordinates of the unmanned agricultural machine during driving, collects the surrounding environment image, and uses the image edge detection technology to extract the edge contour of the pattern in the surrounding environment image, thereby Identify the vertices of the pattern in the surrounding environment image, and use the vertices as feature corner points, and calculate the pixel coordinates of the feature corner points;

The coordinate calibration device is used to correct and calibrate the geodetic coordinates of the unmanned agricultural machine by using the geodetic coordinates and pixel coordinates of the feature corner points.

On the other hand, the embodiment of the present invention also provides a navigation device for unmanned agricultural machinery, as shown in Figure 5, the device includes:

processor 601,

The memory 602 includes program instructions executable by the processor. When the program instructions are executed by the processor, the device for determining the position coordinates of the indoor wireless signal transmission anchor point performs the following operations:

Determine the geodetic coordinates of the characteristic corner points of the graphic signs set on the driving path of the unmanned agricultural machine or on both sides of the driving path;

The geodetic coordinates of the unmanned agricultural machine are located during the driving process of the unmanned agricultural machine, the surrounding environment image is collected, and the edge contour of the pattern in the surrounding environment image is extracted using the image edge detection technology, thereby identifying the pattern in the surrounding environment image vertex, using the vertex as a feature corner point, and calculating the pixel coordinates of the feature corner point;

Using the geodetic coordinates and pixel coordinates of the feature corners, the geodetic coordinates of the unmanned agricultural machine are corrected and calibrated.

An embodiment of the present disclosure provides a computer-readable storage medium, which is characterized in that a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, various processes of the above-mentioned drone positioning and navigation method are implemented, And can achieve the same technical effect, in order to avoid repetition, no more details here.

"First", "second" and similar words used in the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different parts. Words like "comprising" or "comprising" mean that the elements preceding the word cover the elements listed after the word, and do not exclude the possibility of also covering other elements. "Up", "Down", "Left", "Right" and so on are only used to indicate the relative positional relationship. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

All terms (including technical terms or scientific terms) used in the present disclosure have the same meaning as understood by one of ordinary skill in the art to which the present disclosure belongs, unless otherwise specifically defined. It should also be understood that terms defined in, for example, general-purpose dictionaries should be interpreted as having meanings consistent with their meanings in the context of the relevant technology, and should not be interpreted in idealized or extremely formalized meanings, unless explicitly stated herein Defined like this.

Techniques, methods and devices known to those of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, such techniques, methods and devices should be considered part of the description.

Claims (10)

1. A method for unmanned agricultural machinery navigation, the method comprising: Determine the geodetic coordinates of the characteristic corner points of the graphic signs set on the driving path of the unmanned agricultural machine or on both sides of the driving path; The geodetic coordinates of the unmanned agricultural machine are located during the driving process of the unmanned agricultural machine, the surrounding environment image is collected, and the edge contour of the pattern in the surrounding environment image is extracted using the image edge detection technology, thereby identifying the pattern in the surrounding environment image vertex, using the vertex as a feature corner point, and calculating the pixel coordinates of the feature corner point; Using the geodetic coordinates and pixel coordinates of the feature corners, the geodetic coordinates of the unmanned agricultural machine are corrected and calibrated. 2. unmanned agricultural machine navigation method according to claim 1, wherein, determine the feature angle that is arranged on the unmanned agricultural machine travel path or the both sides of travel path of the graphic indicator according to the method described in claim 1, point Geodetic coordinates include: Utilize the RTK high-precision receiver to statically measure and collect the coordinate position of the graphic signboard within a period of time, take the average value of the coordinate positions of the graphic signboard collected during the period of time, and calculate the said graphic signpost Geodetic coordinates of feature corners. 3. The unmanned agricultural machine navigation method according to claim 1, wherein, utilizing image edge detection technology, extracting the edge profile of the pattern in the surrounding environment image comprises: calculating the gradient and angle of the pattern in the surrounding environment image; performing non-maximum suppression on the gradient; The edges of the pattern are connected using double thresholding until the complete edge profile of the pattern is extracted. 4. The unmanned agricultural machine navigation method according to claim 3, wherein calculating the gradient and angle of the pattern in the surrounding environment image comprises: Calculate the gradient and angle of the pattern in the surrounding environment image according to the following formula,

Wherein, x is the abscissa of the pixel of the pattern in the surrounding environment image, y is the ordinate of the pixel of the pattern in the surrounding environment image, f is the gray value of the pattern in the surrounding environment image,

The formed angle matrix is calculated for all pixels of the pattern in the surrounding environment image. 5. The unmanned agricultural machinery navigation method according to claim 3, wherein, carrying out non-maximum suppression to the gradient comprises: Judging whether the gray value of the currently detected point C in the gradient is the largest in the 8-connected neighborhood, if it is the largest, continue to check the gray values of the first intersection point dTmp1 and the second corner point dTmp2 of the gradient direction in the gradient Whether the value is greater than C, if C is greater than the gray value of the first intersection point dTmp1 and the second corner point dTmp2, then consider C to be a maximum value and set the value of C to 1, otherwise consider C to be a non-maximum value and set C The value of is set to 0, and all points C in the gradient are traversed, so as to find the local maximum of the pixel points in the gradient, and complete the non-maximum suppression of the gradient. 6. The unmanned agricultural machine navigation method according to claim 3, wherein, using double thresholds to connect the edges of the pattern until the complete edge profile of the pattern is extracted comprises: Select two thresholds, the two thresholds include a low threshold and a high threshold, the points smaller than the low threshold are considered as false edges and set to 0, and the points greater than the high threshold are considered to be strong edges and set to 1; According to the high-threshold points in the image, they are first connected into contours. When the breakpoint of the contour is reached, the algorithm will look for points that meet the low threshold in the 8 fields of the breakpoint, and then collect new edges based on this point until the entire The image is closed. 7. The unmanned agricultural machinery navigation method according to claim 1, wherein, using the earth coordinates and pixel coordinates of the characteristic corner points, correcting and calibrating the earth coordinates of the unmanned agricultural machinery comprises: Calculate the coordinates of the corrected unmanned agricultural machine according to the following formula,

Among them, p' _j is the geodetic coordinate of the characteristic corner point, p _j is the correction coordinate of the unmanned agricultural machine to be optimized, and z' _ij is the characteristic corner point p' observed by the unmanned agricultural machine at the pose T _i The pixel measurement data generated by _j , eij is an error function, and h(T _i , pj) is a projection function from world coordinates to pixel coordinates. 8. A navigation device for unmanned agricultural machinery, wherein the device comprises: The feature corner geodetic coordinate determining device is used to determine the geodetic coordinates of the feature corner points of the graphic signage set on the driving path of the unmanned agricultural machinery or on both sides of the driving path; The feature corner point pixel coordinate calculation device locates the unmanned agricultural machine to obtain the geodetic coordinates of the unmanned agricultural machine during driving, collects the surrounding environment image, and uses the image edge detection technology to extract the edge contour of the pattern in the surrounding environment image, thereby Identifying the vertices of the pattern in the surrounding environment image, using the vertices as feature corner points, and calculating the pixel coordinates of the feature corner points; The coordinate calibration device is used to correct and calibrate the geodetic coordinates of the unmanned agricultural machine by using the geodetic coordinates and pixel coordinates of the feature corner points. 9. A navigation device for unmanned agricultural machinery, the device comprising: processor, The memory includes program instructions executable by the processor. When the program instructions are executed by the processor, the device for determining the position coordinates of the indoor wireless signal transmission anchor point performs the following operations: Determine the geodetic coordinates of the characteristic corner points of the graphic signs set on the driving path of the unmanned agricultural machine or on both sides of the driving path; The geodetic coordinates of the unmanned agricultural machine are located during the driving process of the unmanned agricultural machine, the surrounding environment image is collected, and the edge contour of the pattern in the surrounding environment image is extracted using the image edge detection technology, thereby identifying the pattern in the surrounding environment image vertex, using the vertex as a feature corner point, and calculating the pixel coordinates of the feature corner point; Using the geodetic coordinates and pixel coordinates of the feature corners, the geodetic coordinates of the unmanned agricultural machine are corrected and calibrated. 10. A computer-readable storage medium, on which a computer program is stored, characterized in that, when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images