Narrow channel inter-column navigation moving method and system

Abstract

The invention relates to a narrow channel inter-column navigation movement method and system, and the method comprises the steps: obtaining the coordinate deviation information of the real-time coordinates of a robot and the coordinates of a target point in real time in the autonomous navigation process of the robot; based on a 3D laser radar carried on the robot, laser radar point cloud data of the stand column are collected, and the inflection point position is calculated; based on the inflection point position, distance deviation information of the robot and the stand column is calculated; constructing an in-place judgment model based on the coordinate deviation information and the distance deviation information; and based on the in-place judgment model, it is judged that the robot moves in place. According to the invention, the reliability of real-time positioning and the accuracy of the parking position can be realized in the inter-column running process of the inspection robot in the column type maintenance channel environment.

Description

Technical Field

[0001] This invention relates to the field of robotics, and in particular to a method and system for navigation and movement between pillars in narrow passages. Background Technology

[0002] One common working scenario for train inspection robots is the column-type maintenance trench. The columns on both sides of the maintenance trench often have uniform and repetitive geometric features, which poses a challenge to traditional robot positioning solutions. The distance between two adjacent columns is relatively narrow, but when performing tasks, the inspection robot often needs to cross the gaps between the columns to reach the adjacent track.

[0003] For column-type maintenance trench environments, the mainstream technical solutions and their limitations in this application scenario are as follows: Based on 2D laser SLAM navigation, 2D laser radar is used to scan the environment to obtain two-dimensional planar point clouds, and a planar map is constructed through algorithms to achieve synchronous positioning and mapping. However, this solution is prone to positioning loss or incorrect matching in long corridors due to feature repetition. Summary of the Invention

[0004] The purpose of this invention is to propose a method and system for navigation and movement between pillars in narrow passages. It utilizes the rich three-dimensional information provided by 3D LiDAR, takes the pillars in the environment as stable and reliable key features, and achieves high-precision positioning by fusing positioning information with 3D LiDAR SLAM positioning through template matching.

[0005] To achieve the above objectives, the present invention provides the following solution: A method for navigation and movement between columns in a narrow passage includes: During the robot's autonomous navigation process, the coordinate deviation information between its own real-time coordinates and the target point coordinates is obtained in real time; Based on the 3D LiDAR mounted on the robot, the LiDAR point cloud data of the column is collected, and the inflection point position is calculated. Based on the inflection point position, calculate the distance deviation information between the robot and the column; Based on the coordinate deviation information and the distance deviation information, a positioning determination model is constructed; Based on the positioning determination model, the robot is determined to have moved into position.

[0006] Optionally, the coordinate deviation information is: x_diff=Xx; y_diff=Yy; Where (x,y) are the robot's real-time coordinates, (X,Y) are the target point coordinates, x_diff is the x-axis coordinate deviation data, and y_diff is the y-axis coordinate deviation data.

[0007] Optionally, the lidar data of the column is collected, and the inflection point position is calculated, including: The original lidar point cloud data is filtered, clustered, and fitted to separate the features of the pillars. Based on the features of separation, a plane fitting algorithm is used to fit the plane equations of two adjacent sides of a single column. Then, the spatial equation of the intersection line is calculated by intersecting the two plane equations. The coordinates projected onto the horizontal plane are the inflection point position. The inflection point position refers to the projection of the intersection line of the two sides of the column onto the horizontal plane.

[0008] Optionally, the distance deviation information includes: lateral distance difference d_diff=Dd and longitudinal distance difference h_diff=Hh; Wherein, the difference in lateral distance between the inflection point and the center of the robot is the difference in lateral distance d between the robot and the column, and the difference in longitudinal distance is the difference in longitudinal distance h between the robot and the column. D is the standard lateral distance between the center of the robot's preset parking position and the column, and H is the standard longitudinal distance between the robot's preset parking position and the column.

[0009] Optionally, the positioning determination model is: d_final=k +(1-k) ; Where k∈[0,1], d_final is the distance to the destination, d_diff is the lateral distance difference, h_diff is the longitudinal distance difference, x_diff is the x-axis coordinate deviation data, and y_diff is the y-axis coordinate deviation data.

[0010] Optionally, determining that the robot has moved into position includes: When the weighted calculated value d_final is less than the set threshold, the robot is determined to have moved into position.

[0011] Optionally, the value of k is set as follows: Let distance_xy = When distance_xy >= 0.1, k = 1; when distance_xy <= 0.05, k = 0; when 0.1 <distance_xy<0.05,k=20 distance_xy-1.

[0012] A narrow-channel inter-pillar navigation and movement system, using the aforementioned narrow-channel inter-pillar navigation and movement method, includes: robot; A real-time coordinate acquisition device for the robot is mounted on the robot. 3D LiDAR; Data processing system.

[0013] The beneficial effects of this invention are as follows: This invention enables the inspection robot to achieve reliable real-time positioning and precise parking position control while moving between columns in a column-type maintenance trench environment. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a schematic flowchart of a narrow channel inter-pillar navigation and movement method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of column detection in a column environment according to an embodiment of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] like Figure 1 As shown, this embodiment proposes a method for navigation and movement between columns in a narrow channel, including: During the robot's autonomous navigation process, the coordinate deviation information between its own real-time coordinates and the target point coordinates is obtained in real time; Based on the 3D LiDAR mounted on the robot, the LiDAR point cloud data of the column is collected, and the inflection point position is calculated. Based on the inflection point position, calculate the distance deviation information between the robot and the column; Based on the coordinate deviation information and the distance deviation information, a positioning determination model is constructed; Based on the positioning judgment model, it is determined that the robot has moved to the positioning position, which means that the robot has reached the expected endpoint position.

[0019] Specifically, in this embodiment, a preset lateral distance D and a preset longitudinal distance H are set as template distances for the robot to traverse the column scene. The preset lateral distance D is the standard lateral distance between the center of the robot's preset parking position and the column, and the preset longitudinal distance H is the standard longitudinal distance between the robot's preset parking position and the column.

[0020] The S1 robot is capable of autonomous navigation, such as through inertial navigation. During the process of moving from the starting point P0 to the expected parking position, i.e. the endpoint (X,Y), the robot can perceive its own real-time coordinates (x,y) in real time. Its own real-time coordinates (x,y) can be obtained through odometry and other means. During this process, the robot's motion control before reaching the endpoint depends on x_diff=Xx and y_diff=Yy. When the S2 robot autonomously moves to a certain position, the 3D LiDAR on the robot begins to acquire the features of the pillars on both sides in front, collects the point cloud of the pillars, and performs real-time detection of the pillar features: first, the features of the pillars are separated by filtering, clustering, and fitting the original point cloud; then, the plane equations of the two adjacent sides of a single pillar are fitted by a plane fitting algorithm combining RANSAC and least squares method; then, the spatial equation of the intersection line is calculated by intersecting the two plane equations, and the coordinates projected onto the horizontal plane are the position of the inflection point. The difference in the lateral distance between the position of the inflection point and the position of the robot center is the lateral distance d of the robot from the pillar, and the difference in the longitudinal distance is the longitudinal distance h of the robot from the pillar. In one embodiment, the 3D LiDAR is set at the center of the robot, with the center of the robot as the origin (0,0), and the x' and y' values ​​of the inflection point coordinates obtained by the 3D LiDAR data are used as the horizontal distance d and the vertical distance h, respectively.

[0021] The 3D LiDAR can also be set at other locations on the robot. Based on the relative position of the 3D LiDAR and the center position of the robot, as well as the coordinates of the inflection point, the lateral distance d and the longitudinal distance h can be calculated.

[0022] S3 sets the lateral distance difference d_diff = Dd and the longitudinal distance difference h_diff = Hh. Based on the weighted average of the four variables x_diff, y_diff, d_diff, and h_diff, a final positioning determination formula is constructed: d_final = k. +(1-k) , where k∈[0,1], let distance_xy = When distance_xy >= 0.1, k = 1; when distance_xy <= 0.05, k = 0; when 0.1 <distance_xy<0.05,k=20 distance_xy-1. The robot is considered to have moved into position when the weighted calculated value d_final is less than a set threshold.

[0023] like Figure 2 As shown, in a specific embodiment, applied to a scenario such as train inspection, the maintenance trench contains four pillars. A 3D LiDAR collects data from pillar 3 (the farthest from the robot's initial position and located to the left front of the robot's forward direction) for the aforementioned analysis process. To avoid interference from pillar 4, the 3D LiDAR data range is filtered, retaining only the 90° range to the left front. The robot starts from the starting position, and the 3D LiDAR collects data from pillar 1. For this data, since the current robot position is too far from the endpoint coordinates, the lateral distance d_diff and longitudinal distance h_diff calculated from the pillar features are not referenced. As the robot travels between pillars 1 and 3, distance_xy gradually decreases from >0.1, and d_final also gradually decreases. When d_final is less than a set threshold, the robot has moved into position.

[0024] The method described in this embodiment can improve the reliability of positioning in narrow passageway and intercolumnar environments.

[0025] First, the inspection robot in this solution employs a precise positioning method that combines 3D laser SLAM with real-time column recognition by a front-mounted 3D laser radar. The core of this method lies in utilizing the rich 3D information provided by the 3D laser radar, treating the columns in the environment as stable and reliable key features, and fusing the positioning information with 3D laser SLAM positioning through template matching to achieve high-precision positioning.

[0026] The inspection robot in this embodiment uses a 16-line LiDAR, IMU, and odom odometry as core sensors. It constructs a point cloud map of the on-site environment through IMU attitude estimation and point cloud feature matching, and then optimizes the map through loop closure detection.

[0027] Real-time detection of column features in the environment involves first separating the column features through filtering, clustering, and fitting of the original point cloud. Then, by calculating the positions of the column edges and inflection points, and combining this with the current coordinates of the LiDAR relative to the robot, the distance and angular deviation between the column and the robot can be calculated. Pre-set distance parameters serve as one of the bases for robot localization. The high-precision, low-drift real-time robot pose is input into the robot's motion controller, compared with the preset target path, and control commands (velocity, angular velocity) are generated to drive the robot precisely along the centerline of the passage, safely completing the task of traversing between columns.

[0028] Figure 2 This diagram illustrates column detection in a column environment. A pre-defined lateral distance D and longitudinal distance H serve as template distances for the robot to traverse the column scene. During the robot's movement from the starting point to the endpoint (X,Y), its real-time coordinates are (x,y). The robot's control before reaching the endpoint depends on x_diff=Xx and y_diff=Yy. Once the 3D LiDAR acquires the column features on both sides and calculates the lateral distance d and longitudinal distance h between the robot and the column, the lateral distance difference d_diff=Dd and the longitudinal distance difference h_diff=Hh are set. A weighted formula for determining the robot's position is constructed based on the four variables x_diff, y_diff, d_diff, and h_diff. When the weighted calculated value is less than a set threshold, the robot is considered to have reached its position.

[0029] This embodiment also proposes a narrow-channel inter-pillar navigation and movement system, which, when applied to the aforementioned narrow-channel inter-pillar navigation and movement method, includes: robot; A real-time coordinate acquisition device for the robot is mounted on the robot. 3D LiDAR; Data processing system.

[0030] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for navigation and movement between pillars in a narrow passage, characterized in that, include: During the robot's autonomous navigation process, the coordinate deviation information between its own real-time coordinates and the target point coordinates is obtained in real time; Based on the 3D LiDAR mounted on the robot, the LiDAR point cloud data of the column is collected, and the inflection point position is calculated. Based on the inflection point position, calculate the distance deviation information between the robot and the column; Based on the coordinate deviation information and the distance deviation information, a positioning determination model is constructed; Based on the positioning determination model, the robot is determined to have moved into position.

2. The method for navigation and movement between pillars in a narrow passage according to claim 1, characterized in that, The coordinate deviation information is: x_diff=Xx; y_diff=Yy; Where (x,y) are the robot's real-time coordinates, (X,Y) are the target point coordinates, x_diff is the x-axis coordinate deviation data, and y_diff is the y-axis coordinate deviation data.

3. The method for navigation and movement between pillars in a narrow passage according to claim 1, characterized in that, Collect lidar data from the pillar and calculate the inflection point location, including: The original lidar point cloud data is filtered, clustered, and fitted to separate the features of the pillars. Based on the features of separation, a plane fitting algorithm is used to fit the plane equations of two adjacent sides of a single column. Then, the spatial equation of the intersection line is calculated by intersecting the two plane equations. The coordinates projected onto the horizontal plane are the inflection point position. The inflection point position refers to the projection of the intersection line of the two sides of the column onto the horizontal plane.

4. The method for navigation and movement between pillars in a narrow passage according to claim 1, characterized in that, The distance deviation information includes: lateral distance difference d_diff=Dd and longitudinal distance difference h_diff=Hh; Wherein, the difference in lateral distance between the inflection point and the center of the robot is the difference in lateral distance d between the robot and the column, and the difference in longitudinal distance is the difference in longitudinal distance h between the robot and the column. D is the standard lateral distance between the center of the robot's preset parking position and the column, and H is the standard longitudinal distance between the robot's preset parking position and the column.

5. The method for navigation and movement between pillars in a narrow passage according to claim 1, characterized in that, The positioning determination model is as follows: d_final=k +(1-k) ; Where k∈[0,1], d_final is the distance to the destination, d_diff is the lateral distance difference, h_diff is the longitudinal distance difference, x_diff is the x-axis coordinate deviation data, and y_diff is the y-axis coordinate deviation data.

6. The method for navigation and movement between pillars in a narrow passage according to claim 5, characterized in that, Determining whether the robot has moved into position includes: When the weighted calculated value d_final is less than the set threshold, the robot is determined to have moved into position.

7. A navigation and movement system between pillars in a narrow passage, characterized in that, The method for navigation and movement between pillars in a narrow passage as described in any one of claims 1 to 6 includes: robot; A real-time coordinate acquisition device for the robot is mounted on the robot. 3D LiDAR; Data processing system.

Images