CN-122024193-A - Fixed wing unmanned aerial vehicle landing navigation method, device, equipment and medium

CN122024193ACN 122024193 ACN122024193 ACN 122024193ACN-122024193-A

Abstract

The application discloses a landing navigation method, device, equipment and medium of a fixed wing unmanned aerial vehicle, and relates to the technical field of unmanned aerial vehicle control, comprising the steps of collecting images of the landing stage of the unmanned aerial vehicle and determining a runway area from the images; the method comprises the steps of carrying out gray processing on an image, extracting candidate line segments in the gray image by using a region growing algorithm, determining a horizon in a forward view angle image, screening the candidate line segments to obtain a screened runway line set, carrying out runway line detection on a runway region to obtain a corresponding runway line detection result, carrying out contrast verification on the line segments in the screened runway line set and the runway line detection result to determine a target runway line, and carrying out landing navigation based on the target runway line, the horizon and camera internal reference pairs of a preset camera. Through screening based on geometric constraint and consistency verification means among detection results of different algorithms, false detection caused by shadows, interference lines and the like is effectively eliminated, and accordingly landing reliability of the unmanned aerial vehicle is guaranteed.

Inventors

CHEN BEI
ZHOU ZEBO
TANG SHIXING
Lin Jiejian
LEI YANG
ZHANG ZIMU
QIN PENG
WANG YISU

Assignees

中航(成都)无人机系统股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260203

Claims (10)

1. The landing navigation method of the fixed wing unmanned aerial vehicle is characterized by comprising the following steps of: Acquiring a forward visual angle image of the unmanned aerial vehicle in a landing stage by using a preset camera on the target fixed wing unmanned aerial vehicle, and determining a runway area from the forward visual angle image based on a target detection network; gray processing is carried out on the forward view angle image so as to obtain a corresponding gray image, a candidate line segment in the gray image is extracted by utilizing a region growing algorithm, and a horizon in the forward view angle image is determined by utilizing a random sampling consistency algorithm; screening the candidate line segments based on the horizon and a preset runway line constraint method to obtain a screened runway line set; Performing runway line detection on the runway area by using an ultrafast runway line detection algorithm to obtain a corresponding runway line detection result, and performing geometric contrast verification on line segments in the screened runway line set and the runway line detection result to determine a target runway line; And determining the attitude angle of the target fixed wing unmanned aerial vehicle relative to the runway based on the target runway line, the horizon line and the camera internal parameters of the preset camera so as to perform landing navigation on the target fixed wing unmanned aerial vehicle.
2. The fixed wing unmanned aerial vehicle landing navigation method of claim 1, wherein before the determining the runway area from the forward view image based on the object detection network, further comprises: acquiring a historical forward view image set, performing data annotation on the historical forward view image set to acquire a corresponding annotated image set, and dividing the annotated image set into a target training set and a target test set; And training a preset initial detection network by using the target training set to obtain a corresponding trained network, testing the detection effect of the trained network by using the target testing set, and determining the trained network as the target detection network if the detection effect of the trained network passes the test.
3. The fixed wing unmanned aerial vehicle landing navigation method of claim 1, wherein the extracting candidate line segments in the grayscale image using a region growing algorithm comprises: Carrying out Gaussian filtering and downsampling on the gray level image, calculating the gradient amplitude and the gradient direction of each pixel point in the corresponding processed image, and sequencing each pixel point according to the gradient amplitude of each pixel point; and selecting seed points from the pixel points according to the corresponding sorting results to perform region growth so as to combine adjacent pixels with consistent gradient directions into the same connected region, and determining the candidate line segments based on gradient distribution of the pixels in each connected region.
4. The fixed wing unmanned aerial vehicle landing navigation method of claim 1, wherein the determining the horizon in the forward view image using a random sampling consistency algorithm comprises: screening a target line segment which is positioned above the runway area and meets the preset line length condition, and determining points on the target line segment as candidate feature points; constructing a plurality of characteristic lines based on the candidate characteristic points, and respectively calculating the target distance from each candidate characteristic point to each characteristic line; determining points with the target distance smaller than a preset distance threshold as internal points, and recording the number of the internal points corresponding to each characteristic straight line respectively; and determining an inner point set corresponding to the characteristic straight line with the largest inner point number as an optimal inner point set, and fitting each point in the optimal inner point set by using a least square method to obtain the horizon.
5. The fixed wing unmanned aerial vehicle landing navigation method of claim 1, wherein the screening the candidate line segments based on the horizon and a preset runway line constraint method comprises: Determining a line length threshold value based on the size information of the runway area, and removing line segments with line lengths smaller than the line length threshold value from the candidate line segments; And determining a slope deviation threshold value based on the slope information of the horizon, and removing a line segment with the absolute value of the difference between the slope and the slope of the horizon smaller than the slope deviation threshold value from the candidate line segments.
6. The fixed wing unmanned aerial vehicle landing navigation method of claim 1, wherein the geometric comparison verification of the line segments in the screened runway line set and the runway line detection result to determine a target runway line comprises: Respectively calculating the angle difference between a first line segment and a corresponding second line segment, wherein the first line segment is a line segment in the screened runway line set, and the second line segment is a line segment corresponding to the runway line detection result; Selecting a plurality of sampling points in a common ordinate range of the first line segment and the corresponding second line segment, respectively calculating the horizontal coordinate difference values of the first line segment and the corresponding second line segment at each sampling point, and determining the average horizontal offset between the first line segment and the corresponding second line segment based on the horizontal coordinate difference values corresponding to each sampling point; And if the angle difference is smaller than a preset angle difference threshold value and the average transverse offset is smaller than a preset offset threshold value, determining the first line segment as the target runway line.
7. The fixed wing unmanned aerial vehicle landing navigation method of claim 1, wherein the determining the attitude angle of the target fixed wing unmanned aerial vehicle relative to the runway based on the target runway line, the horizon line, and the camera references of the preset camera comprises: Determining vanishing points according to the intersection points of the target runway lines on the forward visual angle images, and acquiring yaw angles and pitch angles of the target fixed wing unmanned aerial vehicle based on vanishing point coordinates of the vanishing points and the camera internal parameters; If the runway end line meeting the preset length condition is detected, calculating the roll angle of the target fixed wing unmanned aerial vehicle according to the slope of the runway end line; and if the runway end line meeting the preset length condition is not detected, determining the roll angle of the target fixed wing unmanned aerial vehicle based on the horizon line.
8. A fixed wing unmanned aerial vehicle landing navigation device, comprising: The image acquisition module is used for acquiring a forward visual angle image of the unmanned aerial vehicle in a landing stage by using a preset camera on the target fixed wing unmanned aerial vehicle, and determining a runway area from the forward visual angle image based on a target detection network; The horizon determining module is used for carrying out gray processing on the forward view angle image to obtain a corresponding gray image, extracting candidate line segments in the gray image by using a region growing algorithm, and determining the horizon in the forward view angle image by using a random sampling consistency algorithm; The line segment screening module is used for screening the candidate line segments based on the horizon and a preset runway line constraint method so as to obtain a screened runway line set; The runway line determining module is used for detecting runway lines of the runway area by using an ultrafast lane line detecting algorithm to obtain corresponding runway line detecting results, and performing geometric contrast verification on line segments in the screened runway line set and the runway line detecting results to determine target runway lines; And the attitude angle determining module is used for determining the attitude angle of the target fixed wing unmanned aerial vehicle relative to the runway based on the target runway line, the horizon line and the camera internal parameters of the preset camera so as to perform landing navigation on the target fixed wing unmanned aerial vehicle.
9. An electronic device, comprising: A memory for storing a computer program; a processor for executing the computer program to implement the fixed wing unmanned aerial vehicle landing navigation method of any of claims 1 to 7.
10. A computer readable storage medium for storing a computer program which when executed by a processor implements a fixed wing unmanned aerial vehicle landing navigation method as claimed in any one of claims 1 to 7.

Description

Fixed wing unmanned aerial vehicle landing navigation method, device, equipment and medium Technical Field The invention relates to the technical field of unmanned aerial vehicle control, in particular to a fixed wing unmanned aerial vehicle landing navigation method, a device, equipment and a medium. Background The existing landing navigation method of the fixed wing unmanned aerial vehicle comprises the steps of detecting and restricting a screening runway line through a multi-scale LSD (LINE SEGMENT Detector, namely a line segment detection algorithm), fitting a horizon line for standby through a RANSAC (Random Sample Consensus, namely a random sampling consistency algorithm), and finally carrying out fine detection and outputting navigation parameters through a UFLD (Ultra Fast Lane Detection, namely an ultra-fast lane detection algorithm), so that landing guidance of the unmanned aerial vehicle is carried out. The method still has the problem of higher false detection rate of the runway line under the complex environment (such as interference lines on the periphery of the runway and complex ground texture), so that the landing reliability is reduced, and therefore, how to improve the landing reliability becomes the technical problem to be solved at present. Disclosure of Invention In view of the above, the invention aims to provide a landing navigation method, device, equipment and medium for a fixed wing unmanned aerial vehicle, which can effectively eliminate false detection caused by shadows, interference lines and the like through screening based on geometric constraints and consistency verification means among detection results of different algorithms, thereby ensuring the landing reliability of the unmanned aerial vehicle. The specific scheme is as follows: in a first aspect, the present application provides a fixed wing unmanned aerial vehicle landing navigation method, including: Acquiring a forward visual angle image of the unmanned aerial vehicle in a landing stage by using a preset camera on the target fixed wing unmanned aerial vehicle, and determining a runway area from the forward visual angle image based on a target detection network; gray processing is carried out on the forward view angle image so as to obtain a corresponding gray image, a candidate line segment in the gray image is extracted by utilizing a region growing algorithm, and a horizon in the forward view angle image is determined by utilizing a random sampling consistency algorithm; screening the candidate line segments based on the horizon and a preset runway line constraint method to obtain a screened runway line set; Performing runway line detection on the runway area by using an ultrafast runway line detection algorithm to obtain a corresponding runway line detection result, and performing geometric contrast verification on line segments in the screened runway line set and the runway line detection result to determine a target runway line; And determining the attitude angle of the target fixed wing unmanned aerial vehicle relative to the runway based on the target runway line, the horizon line and the camera internal parameters of the preset camera so as to perform landing navigation on the target fixed wing unmanned aerial vehicle. Optionally, before the determining, based on the object detection network, the runway area from the forward view image, the method further includes: acquiring a historical forward view image set, performing data annotation on the historical forward view image set to acquire a corresponding annotated image set, and dividing the annotated image set into a target training set and a target test set; And training a preset initial detection network by using the target training set to obtain a corresponding trained network, testing the detection effect of the trained network by using the target testing set, and determining the trained network as the target detection network if the detection effect of the trained network passes the test. Optionally, the extracting the candidate line segments in the gray scale image by using a region growing algorithm includes: Carrying out Gaussian filtering and downsampling on the gray level image, calculating the gradient amplitude and the gradient direction of each pixel point in the corresponding processed image, and sequencing each pixel point according to the gradient amplitude of each pixel point; and selecting seed points from the pixel points according to the corresponding sorting results to perform region growth so as to combine adjacent pixels with consistent gradient directions into the same connected region, and determining the candidate line segments based on gradient distribution of the pixels in each connected region. Optionally, the determining the horizon in the forward view image by using a random sampling consistency algorithm includes: screening a target line segment which is positioned above the runway area and meets the preset line length cond