Unmanned aerial vehicle road inspection dynamic obstacle elimination algorithm based on clustering principle

Abstract

The invention discloses an unmanned aerial vehicle road inspection dynamic obstacle elimination algorithm based on a clustering principle, and relates to the field of unmanned aerial vehicles. When an unmanned aerial vehicle carries out road inspection work, environment data of a road to be inspected is extracted through a laser radar and a visible light camera, point cloud data obtained by the laser radar are processed, and the road inspection dynamic obstacle elimination algorithm is obtained; a large amount of redundant information is eliminated, useful point cloud data is extracted, calculation consumption is reduced, dynamic obstacles in an inspection road are eliminated more quickly, fusion operation is performed on two kinds of heterogeneous data, cloud data are converted into pixel points of color texture data, subsequent clustering analysis is facilitated, calculation time cost is reduced through clustering analysis, and the calculation efficiency is improved. The real-time performance is improved, the dynamic point cloud target is identified to facilitate the later elimination work, the point cloud data and the color texture data are spliced and reconstructed to complete the elimination of the dynamic obstacle in the unmanned aerial vehicle routing inspection road, and the unmanned aerial vehicle can conveniently perform the road routing inspection work.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle road inspection, and in particular to a dynamic obstacle elimination algorithm for unmanned aerial vehicle road inspection based on clustering principle. Background Art

[0002] Traditional road inspections usually require manual walking or driving. Faced with a wide range of roads, it is a time-consuming and labor-intensive task to fully check the details. UAVs are unmanned aircraft that are controlled by radio remote control equipment and self-contained program control devices.

[0003] In the prior art, such as the Chinese publication number: CN112141336B, the present invention provides a road safety inspection drone, including a main housing and a swing arm, the end of the swing arm is fixedly connected to a motor, the output end of the motor is fixedly connected to a rotor, the top of the main housing is fixedly connected to a power supply mechanism, and a flight electronic control system is arranged in the main housing. Through the setting of the video transmission system, the drone can transmit the road scene to the traffic police, and the traffic police can perform ticketing operations on the vehicles on the road after judgment, and can also timely query the owner's information according to the vehicle license plate, and remind the owner to move the vehicle in time by calling to avoid traffic accidents and improve the safety of the road. In addition, through the electric telescopic rod, the police ticket PDA, the rubber wheel, the manipulator and the ticket control unit, the ticket can be posted on the illegal vehicle by the manipulator, which improves the efficiency of law enforcement, greatly reduces the workload of the traffic police, is conducive to deterring the owner who wants to violate the law, and further improves the safety of the road.

[0004] In the prior art, such as the Chinese publication number: CN108198417B, the present invention provides a road inspection system based on drones, a road inspection system based on drones, including: a ground control center and a drone; the drone includes a control module, a camera module, a wireless transmission module and a GPS positioning module, wherein: the camera module is used to capture road images; the GPS module is used to accurately locate the drone and obtain real-time positioning information of the drone; the wireless transmission module is used to send the road image and positioning information to the ground control center, and receive the inspection plan sent from the ground control center; the control module is used to control the flight path of the drone according to the inspection plan; the ground control center is used to receive the image and positioning information sent from the drone, and send the inspection plan to the drone. The present invention obtains road images through drones and monitors road conditions in real time, with strong intelligence and high accuracy.

[0005] To sum up, if drones are used for road inspections, it will not only improve work efficiency, but also reduce the waste of human resources. However, during the road inspection process, if drones are conducting inspections on open roads, there are fewer vehicles on the road and the inspection roads are not blocked, the inspection work can be carried out well. However, if there are too many vehicles on the road, the road to be inspected will be blocked, which is not conducive to the drone's road inspection work. Summary of the invention

[0006] The purpose of the present invention is to solve the problem of a dynamic obstacle elimination algorithm for unmanned aerial vehicle road inspection based on the clustering principle. When a unmanned aerial vehicle performs road inspection, the dynamic obstacles on the inspection road are eliminated through the principle of cluster analysis, thereby facilitating the unmanned aerial vehicle to perform road inspection.

[0007] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0008] The dynamic obstacle elimination algorithm for UAV road inspection based on clustering principle includes the following steps:

[0009] S1. The drone obtains environmental data of the road to be inspected through its own environmental perception sensors;

[0010] S2. Process the acquired environmental data, detect targets on the processed data, and extract characteristic points of the inspection road environment;

[0011] S3. Perform cluster analysis on the characteristic points of the inspection road environment to extract dynamic obstacle features;

[0012] S4. Remove dynamic obstacles from the inspection road and align with the original collected data to obtain complete inspection road information.

[0013] Preferably, in step S1, the environmental perception sensor includes a laser radar and a visible light camera, the laser radar is used to obtain point cloud data of the position and reflection intensity of each point in the road to be inspected, and the visible light camera is used to obtain the overall color texture data of the road to be inspected.

[0014] Preferably, in step S2, the data processing includes point cloud data preprocessing and two kinds of heterogeneous data fusion processing, the point cloud data preprocessing is used to remove discrete points, invalid points and outliers in the point cloud data, and extract useful point cloud data, and the two kinds of heterogeneous data fusion processing are used to realize the fusion of the two kinds of heterogeneous data, namely the point cloud data and the color texture data, and eliminate the differences in the forms of the two kinds of data.

[0015] Preferably, in step S2, the point cloud data preprocessing includes the following sub-steps:

[0016] S21. In the point cloud data, a starting point y is selected as the first sampling point, and the set of sampling points is S = {y};

[0017] S22, calculate the distances from all points outside the set point S to the sampling point y, find the point y1 with the farthest distance, and update the sampling set S = {y, y1};

[0018] S23, calculate the distance from all points outside the set S to y1, for any point y i , calculate y i The distance to y2;

[0019] S24, select the maximum arbitrary point y2 and add it to the set S, and update the sampling set S = {y, y1, y2};

[0020] S25, repeating steps S22 to S24 until the required number of samples is reached, thereby completing the extraction of useful point cloud data.

[0021] Preferably, in step S2, the fusion of the two heterogeneous data is to convert the point cloud data into pixel points of the color texture data, and the conversion relationship expression is: Where I is the transformed pixel, S(·) represents the transformation process from a single space to a pixel, T is the translation vector, K is the camera intrinsic parameter matrix, and P is i It is the homogenization of the pixel points.

[0022] Preferably, in step S3, the cluster analysis includes the following sub-steps:

[0023] S31, converting the obtained point cloud data into pixel points of color texture data, and establishing a mapping relationship from three-dimensional points to pixel points;

[0024] S32, calculating the Euclidean distance between the color texture data pixel corresponding to any three-dimensional point of the point cloud data and the edge pixel of the target object to determine the edge of the target point cloud;

[0025] S33, determining the three-dimensional point with the minimum Euclidean distance value and the inner edge of the target point cloud as the center point of the target point cloud;

[0026] S34. After all points in the point cloud data are classified, the point set within the category is used as the segmentation result to complete the cluster analysis.

[0027] Preferably, in step S32, the Euclidean distance formula is: Where, d i is the Euclidean distance of the edge pixel of the target object, X cm and cm are the center point coordinates of the target point cloud, X i andi are the pixel coordinates of any point.

[0028] Preferably, in step S3, the extracting of dynamic obstacle features determines that any target point of the cluster is a dynamic obstacle according to the segmentation result of the cluster analysis, including speed and trajectory information of any target point.

[0029] Preferably, in step S4, the original collected data registration includes splicing and reconstruction of the overall color texture data of the inspection road.

[0030] The beneficial effects of the present invention are:

[0031] When the drone of the present invention performs road inspection, it extracts environmental data of the road to be inspected through a laser radar and a visible light camera, processes the point cloud data acquired by the laser radar, eliminates a large amount of redundant information, extracts useful point cloud data, reduces computing consumption, and eliminates dynamic obstacles in the inspection road more quickly. It also performs a fusion operation on the two heterogeneous data, converts the cloud data into pixel points of the color texture data, and facilitates subsequent clustering analysis. Clustering analysis is used to reduce computing time costs, improve real-time performance, and identify dynamic point cloud targets for later elimination. Through the splicing and reconstruction of point cloud data and color texture data, dynamic obstacles are eliminated on the drone inspection road, which facilitates the drone to perform road inspections and helps to monitor and manage roads more comprehensively and meticulously, thereby improving road safety and reducing waste of human resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 The present invention is a flowchart of the implementation steps of the dynamic obstacle elimination algorithm for unmanned aerial vehicle road inspection based on the clustering principle.

[0033] Figure 2 This is a flow chart of the cluster-free analysis steps of the dynamic obstacle elimination algorithm for unmanned aerial vehicle road inspection based on the clustering principle of the present invention. DETAILED DESCRIPTION

[0034] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further described below in conjunction with specific embodiments, but the following embodiments are only preferred embodiments of the present invention, not all. Based on the embodiments in the implementation mode, other embodiments obtained by those skilled in the art without creative work all belong to the protection scope of the present invention. The experimental methods in the following embodiments, unless otherwise specified, are conventional methods, and the materials, reagents, etc. used in the following embodiments, unless otherwise specified, can all be obtained from commercial channels.

[0035] Combine the following Figure 1 and Figure 2 The present invention is further described.

[0036] The dynamic obstacle elimination algorithm for UAV road inspection based on clustering principle includes the following steps:

[0037] S1. The drone obtains environmental data of the road to be inspected through its own environmental perception sensors.

[0038] The present invention is further configured as follows: the environmental perception sensor includes a laser radar and a visible light camera, the laser radar is used to obtain point cloud data of the position and reflection intensity of each point in the road to be inspected, and the visible light camera is used to obtain the overall color texture data of the road to be inspected.

[0039] S2. Process the acquired environmental data, detect targets on the processed data, and extract feature points of the inspection road environment.

[0040] The present invention is further configured as follows: the data processing includes point cloud data preprocessing and two kinds of heterogeneous data fusion processing. The point cloud data preprocessing is used to remove discrete points, invalid points and outliers in the point cloud data, and extract useful point cloud data. The point cloud data collected by the laser radar usually contains up to 100,000 data points. Directly inputting all the collected point cloud data will result in a huge amount of calculation. The point cloud data preprocessing can eliminate a large amount of redundant information and reduce computing consumption. The two kinds of heterogeneous data fusion processing are used to realize the fusion of the two kinds of heterogeneous data, the point cloud data and the color texture data, and eliminate the differences in the forms of the two data.

[0041] The present invention is further configured as follows: the point cloud data preprocessing comprises the following sub-steps:

[0042] S21. In the point cloud data, a starting point y is selected as the first sampling point, and the set of sampling points is S = {y};

[0043] S22, calculate the distances from all points outside the set point S to the sampling point y, find the point y1 with the farthest distance, and update the sampling set S = {y, y1};

[0044] S23, calculate the distance from all points outside the set S to y1, for any point y i , calculate y i The distance to y2;

[0045] S24, select the maximum arbitrary point y2 and add it to the set S, and update the sampling set S = {y, y1, y2};

[0046] S25, repeating steps S22 to S24 until the required number of samples is reached, thereby completing the extraction of useful point cloud data.

[0047] The present invention is further configured as follows: the fusion of the two heterogeneous data is to convert the point cloud data into pixel points of the color texture data, and the conversion relationship expression is: Where I is the transformed pixel, S(·) represents the transformation process from a single space to a pixel, T is the translation vector, K is the camera intrinsic parameter matrix, and P is i To homogenize the pixel points, it can avoid the large difference in data form between point cloud data and color texture data due to the different sampling data of lidar and visible light camera, which affects the subsequent elimination of dynamic obstacles.

[0048] S3. Perform cluster analysis on the characteristic points of the inspection road environment to extract dynamic obstacle features.

[0049] The present invention is further configured as follows: cluster analysis comprises the following sub-steps:

[0050] S31, converting the obtained point cloud data into pixel points of color texture data, and establishing a mapping relationship from three-dimensional points to pixel points;

[0051] S32, calculating the Euclidean distance between the color texture data pixel corresponding to any three-dimensional point of the point cloud data and the edge pixel of the target object to determine the edge of the target point cloud;

[0052] S33, determining the three-dimensional point with the minimum Euclidean distance value and the inner edge of the target point cloud as the center point of the target point cloud;

[0053] S34. After all points in the point cloud data are classified, the point set within the category is used as the segmentation result to complete the cluster analysis.

[0054] The present invention is further configured as follows: the Euclidean distance formula is: Where, d i is the Euclidean distance of the edge pixel of the target object, X cm and cm are the center point coordinates of the target point cloud, X i and i are the pixel coordinates of any point.

[0055] The present invention is further configured as follows: the extraction of dynamic obstacle features determines that any target point of the cluster is a dynamic obstacle according to the segmentation result of the cluster analysis, including speed and trajectory information of any target point.

[0056] S4. Remove dynamic obstacles from the inspection road and align with the original collected data to obtain complete inspection road information.

[0057] The present invention is further configured such that: the original collected data registration includes the splicing and reconstruction of the overall color texture data of the inspection road.

[0058] To sum up, when the drone of the present invention performs road inspection, it uses laser radar and visible light camera to extract environmental data of the road to be inspected, processes the point cloud data obtained by the laser radar, eliminates a large amount of redundant information, extracts useful point cloud data, reduces computing consumption, and eliminates dynamic obstacles in the inspection road more quickly. It also performs a fusion operation on the two heterogeneous data, converts the cloud data into pixel points of the color texture data, and facilitates subsequent clustering analysis. Clustering analysis is used to reduce computing time costs, improve real-time performance, and identify dynamic point cloud targets for later elimination. Through the splicing and reconstruction of point cloud data and color texture data, dynamic obstacles are eliminated on the drone inspection road, which facilitates the drone to perform road inspections and helps to monitor and manage roads more comprehensively and meticulously, thereby improving road safety and reducing waste of human resources.

[0059] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. The dynamic obstacle elimination algorithm for UAV road inspection based on clustering principle is characterized by: The following steps are involved: S1. The drone obtains environmental data of the road to be inspected through its own environmental perception sensors; S2. Process the acquired environmental data, detect targets on the processed data, and extract characteristic points of the inspection road environment; S3, cluster analysis is performed on the characteristic points of the inspection road environment to extract dynamic obstacle features; S4. Remove dynamic obstacles from the inspection road and align with the original collected data to obtain complete inspection road information.

2. The dynamic obstacle elimination algorithm for UAV road inspection based on clustering principle according to claim 1 is characterized in that: In step S1, the environment perception sensor includes a laser radar and a visible light camera, the laser radar is used to obtain point cloud data of the position and reflection intensity of each point in the road to be inspected, and the visible light camera is used to obtain the overall color texture data of the road to be inspected.

3. The dynamic obstacle elimination algorithm for UAV road inspection based on clustering principle according to claim 1 is characterized in that: In step S2, the data processing includes point cloud data preprocessing and two kinds of heterogeneous data fusion processing. The point cloud data preprocessing is used to remove discrete points, invalid points and outliers in the point cloud data and extract useful point cloud data. The two kinds of heterogeneous data fusion processing are used to realize the fusion of the two kinds of heterogeneous data, namely the point cloud data and the color texture data, and eliminate the differences in the forms of the two data.

4. The dynamic obstacle elimination algorithm for unmanned aerial vehicle road inspection based on clustering principle according to claim 3 is characterized in that: In step S2, the point cloud data preprocessing includes the following sub-steps: S21. In the point cloud data, a starting point y is selected as the first sampling point, and the set of sampling points is S = {y}; S22, calculate the distances from all points outside the set point S to the sampling point y, find the point y1 with the farthest distance, and update the sampling set S = {y, y1}; S23, calculate the distance from all points outside the set S to y1, for any point y i , calculate y i The distance to y2; S24, select the maximum arbitrary point y2 and add it to the set S, and update the sampling set S = {y, y1, y2}; S25, repeating steps S22 to S24 until the required number of samples is reached, thereby completing the extraction of useful point cloud data.

5. The dynamic obstacle elimination algorithm for UAV road inspection based on clustering principle according to claim 3 is characterized in that: In step S2, the fusion of the two heterogeneous data is to convert the point cloud data into the pixel points of the color texture data, and the conversion relationship expression is: Where I is the transformed pixel, S(·) represents the transformation process from a single space to a pixel, T is the translation vector, K is the camera intrinsic parameter matrix, and P is i It is the homogenization of the pixel points.

6. The dynamic obstacle elimination algorithm for UAV road inspection based on clustering principle according to claim 1 is characterized in that: In step S3, the cluster analysis includes the following sub-steps: S31, converting the obtained point cloud data into pixel points of color texture data, and establishing a mapping relationship from three-dimensional points to pixel points; S32, calculating the Euclidean distance between the color texture data pixel corresponding to any three-dimensional point of the point cloud data and the edge pixel of the target object to determine the edge of the target point cloud; S33, determining the three-dimensional point with the minimum Euclidean distance value and the inner edge of the target point cloud as the center point of the target point cloud; S34. After all points in the point cloud data are classified, the point set within the category is used as the segmentation result to complete the cluster analysis.

7. The dynamic obstacle elimination algorithm for UAV road inspection based on clustering principle according to claim 6 is characterized in that: In step S32, the Euclidean distance formula is: Where, d i is the Euclidean distance of the edge pixel of the target object, X cm and cm are the center point coordinates of the target point cloud, X i and i are the pixel coordinates of any point.

8. The dynamic obstacle elimination algorithm for UAV road inspection based on clustering principle according to claim 1 is characterized in that: In step S3, the dynamic obstacle feature extraction determines that any target point of the cluster is a dynamic obstacle according to the segmentation result of the cluster analysis, including the speed and trajectory information of any target point.

9. The dynamic obstacle elimination algorithm for UAV road inspection based on clustering principle according to claim 1 is characterized in that: In step S4, the original collected data registration includes splicing and reconstruction of the overall color texture data of the inspection road.

Images