A large target accurate positioning method based on multi-line laser radar

Abstract

The application discloses a kind of big target accurate positioning method based on multi-line laser radar, comprising the following steps: S1, with irregular three-dimensional target as an example, using two multi-line laser radar, vertical installation is different, make up the problem of multi-line laser radar in vertical direction is sparse, realize the accurate positioning of target in three-dimensional space coordinate, scanning target obtains the point cloud data of target.The application is used in the calculation target actual position and correct position deviation, standard shape template is made, only the position information of real-time point cloud is kept, and the accuracy and stability of the calculation coordinate and deviation are improved by using standard template instead of real-time point cloud.RANSAC plane fitting, ICP point cloud registration and other methods are used in the point cloud processing process, such methods are adaptable to target point clouds of different qualities, improve the stability in the process of point cloud processing, and the final positioning accuracy can be controlled within 20mm.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of large target accurate positioning, and in particular to a large target accurate positioning method based on a multi-line laser radar. BACKGROUND

[0002] In the field of industrial production and logistics transportation, the use of visual technology to realize the identification and positioning of targets is an important part of the automation process. The main visual sensors at present include 2D cameras, 3D cameras, multi-line laser radars, etc., and the identification and positioning of targets are realized through the processing of 2D or 3D images obtained by the sensors. The 2D camera can obtain texture information, but it is easily affected by light, bad weather, changes in shooting angle, etc., resulting in low positioning accuracy or failure to complete the task. The 3D camera, multi-line laser radar and other three-dimensional sensors cannot obtain the texture information of the target surface, but have good anti-interference ability, have low requirements for the working environment, and can obtain three-dimensional information and spatial coordinates of the target, and are suitable for positioning targets with obvious three-dimensional features in outdoor working environments. However, the field of view of the 3D camera is generally small, and for large, distant and large scanning range targets, the multi-line laser radar is more suitable.

[0003] The laser radar is a radar system that uses laser beams to detect the position, speed and other characteristic quantities of the target. The data obtained by the laser radar is point cloud data. Using the point cloud data obtained by the laser radar to identify and position the target is one of the common positioning methods. The laser radar collects data from a multi-line (16 lines, 32 lines or 64 lines) laser beam rotating around a target area to accurately model the three-dimensional space, so as to determine the accurate position, size and attitude of a three-dimensional object in the laser radar coordinate system. In order to realize the application on large machinery or to compensate for the sparsity of vertical point clouds, multiple laser radars are usually used together. In this process, the point cloud data of the laser radar is used to realize the identification and positioning of the target, and therefore, a large target accurate positioning method based on a multi-line laser radar is proposed to solve the above problems. SUMMARY

[0004] The purpose of the present application is to propose a large target accurate positioning method based on a multi-line laser radar to solve the problems of outdoor complex weather interference, lack of clear texture on the target surface, and requirement of large identification range. The method realizes the accurate positioning of targets in outdoor all-weather, large range.

[0005] In order to achieve the above purpose, the present application adopts the following technical scheme:

[0006] A large target accurate positioning method based on a multi-line laser radar, comprising the following steps:

[0007] S1, taking an irregular three-dimensional target as an example, two multi-line laser radars are installed vertically on different planes to make up for the sparseness of the multi-line laser radar in the vertical direction, realize accurate positioning of the coordinates of the target in three-dimensional space, and scan the target to obtain point cloud data of the target;

[0008] S2, the original point cloud scanned by the two laser radars is sequentially subjected to target region framing, bearing surface point cloud removal and discrete point filtering to obtain pure and complete target point cloud;

[0009] S3, edge extraction operation is performed on the point cloud to obtain edge information of the target point cloud;

[0010] S4, to realize the offset of the actual position of the target from the correct position, a target positioning template is made, and the following methods are adopted:

[0011] First, the known size data of the target is used to generate standard and uniform target edge point cloud, and the standard target edge point cloud at the template position is obtained by point cloud registration method;

[0012] Second, the registration process uses the 4PCS method for coarse registration, and then uses the iterative closest point (ICP) method for fine registration;

[0013] S5, the standard point cloud at the actual position of the target is obtained by the point cloud registration method, the input is the standard target edge point cloud at the actual position and the template point cloud, the centroid coordinates of the two groups of point clouds are obtained, and the offset of the actual position of the target from the correct position is calculated through the centroid coordinates. Compared with the prior art, the present application can obtain accurate point cloud data by using two multi-line laser radars installed vertically on different planes, remove discrete points, improve the quality of target point cloud, and help improve the accuracy and stability of coordinate calculation and offset. In addition, RANSAC plane fitting, ICP point cloud registration and other methods are used in the point cloud processing process to improve the positioning accuracy.

[0014] Preferably, the two multi-line laser radars are installed vertically on different planes, and the two multi-line laser radars are time-synchronized to obtain simultaneous frame point cloud data of the two multi-line laser radars, and the two groups of point cloud data are unified to a coordinate system through calibration of the two multi-line laser radars to obtain grid-shaped point cloud data of the target, thereby improving the positioning accuracy. Further, by using the vertical installation method and synchronizing the operation of the installed multi-line laser radars, the quality of the point cloud data is improved to provide accurate data support for improving the positioning accuracy.

[0015] Preferably, the edge information of the point cloud data of the target can be used as a positioning standard, which can greatly improve the calculation efficiency, reduce resource consumption, and use a normal-based point cloud boundary extraction method for the complete target point cloud to obtain the edge point cloud of the target point cloud. Further, the efficiency of subsequent calculations is improved by using the edge information of the point cloud data as a positioning standard, which facilitates rapid and accurate edge point cloud.

[0016] Preferably, the template point cloud generated by the target size data generation standard has the advantages of uniform distribution and stable shape compared with the actual scanned point cloud, which can improve the accuracy and stability of the offset calculation result. Further, by improving the quality of the template point cloud, accurate data support is provided for subsequent improvement of the accuracy and stability of the offset calculation result.

[0017] Compared with the prior art, the beneficial effects of the present application are:

[0018] 1. The present application has reasonable concept, and can automatically and standardize the positioning of large targets in complex environment, effectively improving the accuracy and efficiency of positioning. It effectively solves the problems of outdoor complex weather interference, target surface without clear texture, large recognition range, etc.

[0019] 2. The present application proposes a perfect positioning process for accurate positioning of large targets. For complex working environment and larger working range, the present application uses a multi-line laser radar to scan the target. The multi-line laser radar is not affected by light, has high resolution, strong anti-interference ability, rich information acquisition, and centimeter-level precision, which is beneficial to the identification and positioning of the target.

[0020] In the calculation of the actual position and correct position offset of the target, the present application uses a standard shape template, only retains the position information of the real-time point cloud, uses a standard template instead of real-time point cloud, improves the calculation of coordinates and offset, and improves the accuracy and stability. In the point cloud processing process, RANSAC plane fitting, ICP point cloud registration and other methods are used. Such methods are adaptive to point clouds of different qualities, improve the stability in the point cloud processing process, and the final positioning accuracy can be controlled within 20mm. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 The flowchart of the accurate positioning method of the large target of the present application;

[0022] Figure 2 The radar and target position relationship diagram of the accurate positioning method of the large target of the present application. DETAILED DESCRIPTION

[0023] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0024] The present invention will be further explained below with reference to specific implementation methods: This example provides a method for precise positioning of large targets based on multi-line lidar, such as... Figure 1 As shown, this embodiment uses two lidars to locate an irregular three-dimensional target and outputs the offset between the target's actual position and the correct position. Specifically, it includes the following steps:

[0025] (1) The two lidars are installed perpendicularly to each other to compensate for the sparseness of the multi-line lidars in the vertical direction, so as to achieve accurate positioning of the target's coordinates in three-dimensional space. The positional relationship between the lidar and the target is as follows: Figure 2 As shown. The lidar is activated, time synchronization is performed, and point cloud data from both lidars at the same time frame is acquired and stored. Through calibration of the two lidars, the two sets of point cloud data are unified into a single coordinate system to obtain the target's grid-like point cloud data.

[0026] (2) Preprocess the raw point cloud data obtained in step (1) above to obtain the feature data of the target point cloud. Here, the selected feature is the edge data of the target point cloud. The specific process is as follows:

[0027] (2.1) Remove background point cloud: The original point cloud is segmented, and the target area is selected using the CropBox algorithm in the PCL point cloud library. Useless point cloud areas are filtered out, where the threshold is determined by the spatial positional relationship between the target and the radar. The processed point cloud data of the target area is obtained.

[0028] (2.2) The point cloud obtained in step (2.1) is fitted to a plane using the RANSAC algorithm to obtain the point cloud set of the target bearing surface. By segmenting the point cloud and removing the bearing surface point cloud, a clean and complete target point cloud is obtained; a schematic diagram of the bearing surface is shown below. Figure 2 As shown.

[0029] (2.3) Use a statistical filtering algorithm to filter out discrete points while retaining complete edge information. (2.4) For the complete target point cloud data, use a normal-based point cloud boundary extraction method to obtain the real-time target point cloud edge information PointCloud_R.

[0030] (3) To achieve the offset between the actual position and the correct position of the output target, a target positioning template is created. The steps are as follows:

[0031] (3.1) Using the known size data of the target, starting from the origin of the point cloud coordinate system, standard and uniform target edge point clouds are generated sequentially, which will be referred to as standard target edge point cloud PointCloud_N.

[0032] (3.2) Place the target at the template position, which is the correct position set. Use a lidar to acquire point cloud data, and through the processing in step (2), obtain the target edge point cloud PointCloud_T at the template position.

[0033] (3.3) Through point cloud registration, PointCloud_T is used as the target point cloud and PointCloud_N is used as the input point cloud to obtain the standard target edge point cloud PointCloud_TN at the template position, which is used as the template for calculating the offset. The registration process uses the 4PCS method for coarse registration and then the Iterative Nearest Neighbor (ICP) method for fine registration.

[0034] (4) By point cloud registration, PointCloud_R is used as the target point cloud and PointCloud_TN is used as the input point cloud to obtain the standard target edge point cloud PointCloud_RN of the actual target position.

[0035] (5) Calculate the centroid coordinates of PointCloud_RN and PointCloud_TN respectively. and Since the relative positions of the points inside the two point clouds are exactly the same, but their overall spatial positions are different, the centroid coordinates can reflect the spatial positional relationship between the two point clouds.

[0036] (6) Using the coordinates of the two centroids as input, the formula for calculating the offset between the actual and correct positions of the target is as follows:

[0037]

[0038] This invention proposes a comprehensive positioning process for precise positioning of large targets. For complex working environments and large working areas, this invention employs a multi-line lidar to scan the target. Multi-line lidar is unaffected by lighting conditions, has extremely high resolution, strong anti-interference capabilities, acquires abundant information, and achieves centimeter-level accuracy, which is beneficial for target identification and positioning. In calculating the offset between the actual and correct target positions, this invention uses standard shape templates and employs methods such as RANSAC plane fitting and ICP point cloud registration, achieving a final positioning accuracy within 20mm.

[0039] This invention is logically conceived and can automate and standardize the efficient localization of large targets in complex environments, effectively improving the accuracy and efficiency of localization. It effectively solves problems such as interference from complex outdoor weather, unclear target surface texture, and large recognition range.

[0040] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for precise localization of large targets based on multi-line lidar, characterized in that, Includes the following steps: S1. Taking an irregular three-dimensional target as an example, two multi-line lidars are installed vertically on opposite sides to make up for the problem of sparseness of multi-line lidars in the vertical direction, so as to achieve accurate positioning of the target's coordinates in three-dimensional space and scan the target to obtain the target's point cloud data. S2. The original point cloud scanned by the two lidars is sequentially processed by selecting the target area, removing the point cloud on the bearing surface, and filtering out discrete points to obtain a clean and complete target point cloud; S3. Perform edge extraction on the point cloud to obtain the edge information of the target point cloud; S4. To achieve the offset between the actual position and the correct position of the output target, a target positioning template is created using the following method: First: Using the known size data of the target, generate a standard, uniform target edge point cloud. Then, obtain the standard target edge point cloud with the correct position through point cloud registration, and use it as a template. Second: The registration process uses the 4PCS method for coarse registration, and then uses the Iterative Nearest Neighbor (ICP) method for fine registration; S5. Obtain the target edge point cloud at the actual position of the target through point cloud registration. The input is the target edge point cloud at the actual position and the template point cloud. Obtain the centroid coordinates of the two sets of point clouds. Calculate the offset between the actual position and the correct position of the target through the centroid coordinates.

2. The method for precise positioning of large targets based on multi-line lidar according to claim 1, characterized in that: Two multi-line lidars are installed vertically on opposite sides. The two lidars are synchronized in time to acquire point cloud data from both lidars at the same time frame. By calibrating the two lidars, the two sets of point cloud data are unified into a coordinate system to obtain the grid-like point cloud data of the target, thereby improving the positioning accuracy.

3. The method for precise positioning of large targets based on multi-line lidar according to claim 1, characterized in that: Using the edge information of the target's point cloud data as a localization standard, a point cloud boundary extraction method based on normals is used to extract the edge point cloud of the target point cloud.

Images