CN-122008196-A - Intelligent path planning method for robot to work outside object

CN122008196ACN 122008196 ACN122008196 ACN 122008196ACN-122008196-A

Abstract

The application discloses an intelligent path planning method for robot to work outside an object, which relates to the technical field of robot automation, and comprises the steps of first and first processing path planning, second and processing effect evaluation, third and second processing scheme analysis, wherein the method comprises the steps of dividing the outer surface of a target object into areas and obtaining vertical visual angle depth images of each area, obtaining the optimal slicing direction and optimal slicing width of each area through a neural network technology, and analyzing the slice following paths of the areas according to the optimal slice direction and the optimal slice width, further analyzing the primary processing following paths of the outer surface of the target object, processing the outer surface of the target object once according to the primary processing paths, analyzing the processing effect after the processing is finished, and analyzing the secondary processing scheme of the outer surface of the target object based on the processing effect so as to obtain the slice effect with higher precision and better quality.

Inventors

TAN TIAN
YANG WENFENG
TAN ZHONGXIAN
LI SHAOLONG
YANG XIAOQIANG
YANG BOWEN
Ye Ti
LI JUNCHAO

Assignees

中国民用航空飞行学院

Dates

Publication Date: 20260512
Application Date: 20260126

Claims (10)

1. An intelligent path planning method for robot to work outside an object, comprising the following steps: step one, primary processing path planning, namely analyzing a primary follow-up processing path of the outer surface of a target object; analyzing the processing effect of the robot on the target object through the primary following processing path, and further judging whether the target object needs secondary processing or not; and thirdly, analyzing a secondary processing scheme, namely analyzing the primary processing effect of the target object to obtain the secondary processing scheme.
2. The intelligent path planning method for robot to object external operation according to claim 1, wherein the analysis of the primary follower processing path of the external surface of the target object is as follows: S1, carrying out region segmentation on the outer surface of a target object based on a normal vector of the outer surface of the target object, and obtaining vertical visual angle depth images corresponding to the regions; s2, outputting the optimal slicing direction and the optimal slicing width of each region by using the vertical visual angle depth image of each region as input through a neural network technology; and S3, carrying out point cloud slicing on each region based on the optimal slicing direction and the optimal slicing width to obtain the contour based on the point cloud slicing of each region, generating a slice follow-up path of each region based on the contour and the normal vector of the point cloud slicing of each region, and analyzing the slice follow-up path sequence of each region to obtain the primary follow-up processing path of the outer surface of the target object.
3. The intelligent path planning method for robot to object external operation according to claim 2, wherein the area segmentation of the external surface of the object is performed based on the normal vector of the external surface of the object, and the vertical view depth image corresponding to each area is obtained, and the specific segmentation and obtaining process is as follows: a1, any point on the outer surface of the target object is acquired and is recorded as Point and acquire and connect Any point with the same distance and adjacent points are recorded as Each adjacent point of the points to be acquired Dots and dots Each adjacent point of the points is arbitrarily connected to construct Each plane in which a point lies is measured using a three-dimensional laser scanner The normal vector of each plane where the point is located is obtained Constructing a region of the outer surface of the target object in each plane with the same normal vector in each plane where the point is located, and accordingly obtaining each region of the outer surface of the target object; A2, acquiring a vertical view depth image of a certain area on the outer surface of the target object along the normal vector direction of the area by using a depth camera, and accordingly acquiring vertical view depth images corresponding to all areas on the outer surface of the target object.
4. The intelligent path planning method for the robot to work outside the object according to claim 2, wherein the vertical view depth image of each area is taken as an input, and the specific process is as follows: acquiring the value of each pixel point from the vertical view depth image of each region and recording the value as Wherein Corresponding numbers of the vertical view depth images of the areas, , Is any integer greater than 2, Is the corresponding number of each pixel point in the vertical visual angle depth image, , For any integer greater than 2, according to the calculation formula: normalizing the vertical view depth image of each region, wherein Is the first The first of the vertical view depth images of the individual regions The input value of the individual pixel points, Represent the first The pixel point minimum of the vertical view depth image of each region, Represent the first The pixel point maximum of the vertical view depth image of each region.
5. The intelligent path planning method for the robot to perform the operation outside the object according to claim 2, wherein the performing the point cloud slicing on each area based on the optimal slicing direction and the optimal slicing width, and obtaining the contour based on the point cloud slicing on each area comprises the following specific steps: b1, establishing a three-dimensional rectangular coordinate system by taking any vertex of the target object as an origin, wherein The axis is in the horizontal direction, The axis is in the vertical direction, The axes are parallel to the space height, the positive direction of each coordinate axis is determined by using a right hand rule, the preset length of the coordinate axis is set to represent a unit length, and then three-dimensional coordinate data of point clouds of each region are obtained; b2, respectively marking the optimal slicing direction and the optimal slicing width of each area on the outer surface of the target object as And Based on three-dimensional coordinate data of point clouds of all areas, according to planes Slicing each regional point cloud, wherein For the number corresponding to each point in the point cloud, , Is any integer greater than 2, Represent the first Point cloud of individual region Cutting coordinate values of the points according to the optimal slicing direction of the outer surface of the target object According to the plane And constructing a slice follower path of each area on the outer surface of the target object.
6. The intelligent path planning method for robot to object external operation according to claim 5, wherein the analyzing the slice follower path sequence of each region further obtains a primary follower processing path of the external surface of the target object comprises the following specific steps: Comparing the optimal slice widths of all the areas of the target object with each other, acquiring an area corresponding to the maximum slice width, marking the area as a first slice area, further comparing the optimal slice widths of all the adjacent areas of the first slice area with each other, acquiring an area corresponding to the maximum slice width of all the adjacent areas of the first slice area, marking the area as a second slice area, and accordingly acquiring the slice sequence of all the areas on the outer surface of the target object, and further constructing a primary follow-up processing path of the outer surface of the target object.
7. The intelligent path planning method for robot to work outside object according to claim 6, wherein the analyzing the processing effect of the robot to the target object through the one-time follow-up processing path comprises the following steps: Acquiring absolute values of differences between actual slice widths and optimal slice widths of all areas of the outer surface of the target object based on primary processing effects of the outer surface of the target object, comparing the absolute values of differences between the actual slice widths and the optimal slice widths with a preset slice width error threshold, and when the absolute values of differences between the actual slice widths and the optimal slice widths of certain areas are larger than the preset slice width error threshold, marking primary processing of the outer surface of the target object by the robot as unqualified slice widths, otherwise marking primary processing of the outer surface of the target object by the robot as qualified slice widths; Acquiring deviation angle values of the actual slicing direction and the optimal slicing direction of each area of the outer surface of the target object by the robot based on the primary processing effect of the outer surface of the target object, comparing the deviation angle values of the actual slicing direction and the optimal slicing direction with a preset threshold value of the deviation angle values of the slicing direction, and when the deviation angle values of the actual slicing direction and the optimal slicing direction of a certain area are larger than the preset threshold value of the deviation angle values of the slicing direction, marking the primary processing of the outer surface of the object by the robot as unqualified slicing direction, otherwise marking the primary processing of the outer surface of the object by the robot as qualified slicing direction; When the slice width and the slice direction of the robot on the outer surface of the target object are qualified, slice width and slice direction deviation angle values of all areas are obtained, the quality evaluation coefficient of the robot on the outer surface of the target object is obtained through analysis, the quality evaluation coefficient of the robot on the outer surface of the target object is compared with a preset quality evaluation coefficient threshold value, when the quality evaluation coefficient of the robot on the outer surface of the target object is larger than or equal to the preset quality evaluation coefficient threshold value, the quality of the robot on the outer surface of the target object is qualified, and otherwise, the quality of the robot on the outer surface of the target object is unqualified.
8. The intelligent path planning method for the robot to work outside the object according to claim 7, wherein the analysis obtains a primary processing quality evaluation coefficient of the robot to the outer surface of the target object, and the specific analysis process is as follows: The slice width and the slice direction deviation angle values of the areas are respectively recorded as And According to the calculation formula: analyzing to obtain one-time processing evaluation coefficient of the robot to the outer surface of the target object Wherein And The weight factor corresponding to the slice width and the weight factor corresponding to the slice direction deviation angle when the robot processes the outer surface of the target object once are respectively shown.
9. The intelligent path planning method for the robot to perform the operation outside the object according to claim 8, wherein the determining whether the target object needs the secondary processing comprises the following steps: when the slice width of the robot for primary processing on the outer surface of the target object is unqualified, judging that the target object needs secondary processing; when the slicing direction of the robot for primary processing on the outer surface of the target object is unqualified, judging that the target object needs secondary processing; and when the machining quality of the robot for primary machining of the outer surface of the target object is unqualified, judging that the target object needs secondary machining.
10. The intelligent path planning method for the robot to work outside the object according to claim 9, wherein the secondary processing scheme is obtained according to the primary processing effect analysis of the target object, and the specific process is as follows: For unqualified slicing directions of the robot on the outer surface of the target object, acquiring deviation angle values of slicing directions and optimal slicing directions of all areas with unqualified slicing directions, sorting the deviation angle values of the slicing directions and the optimal slicing directions of all areas with unqualified slicing directions according to the sequence from big to small, obtaining secondary processing tracks of the robot on all areas on the outer surface of the target object according to the sorting result, and performing similar analysis to obtain secondary processing tracks of the robot on all areas on the outer surface of the target object when the processing quality of the robot on the outer surface of the target object is unqualified; For unqualified machining quality of the robot on the outer surface of the target object, analyzing a primary machining quality evaluation coefficient of the robot on each region of the outer surface of the target object according to the slice width and slice direction deviation angle values of each region of the outer surface of the target object, sorting the primary machining quality evaluation coefficients of the robot on each region of the outer surface of the target object from small to large, and obtaining secondary machining tracks of the robot on each region of the outer surface of the target object according to sorting results; and obtaining a secondary processing track when the slice width, the slice direction and the slice quality of the primary processing of the robot on the outer surface of the target object are unqualified according to the secondary processing track, and carrying out secondary processing on the outer surface of the target object according to the secondary processing track.

Description

Intelligent path planning method for robot to work outside object Technical Field The application relates to the technical field of robot automation, in particular to an intelligent path planning method for robot operation outside an object. Background In order to realize the technologies of automatic cleaning, spraying, polishing, grinding and the like of a robot on the surface of a workpiece, the problems of path and motion planning of a mechanical arm on the surface of the workpiece are inevitably involved, the motion path of the mechanical arm can be planned by dividing the surface area of the workpiece and combining specific processing requirements at present, but the processing precision is still yet to be improved, so the application provides an intelligent path planning method for the robot to follow-up operation of the outer surface of an object. The intelligent cleaning robot path planning method and system disclosed in the patent application of the invention with the publication number of CN113012149A in the prior art comprises the steps of collecting image data of characteristic parts and stain areas of an object to be cleaned by a depth camera, preprocessing the image, identifying and dividing the image data of the stain areas of the object to be cleaned by a convolutional neural network, and planning the cleaning path of the cleaning robot by combining prior information of the shape and the size of the object to be cleaned. The invention can utilize image information to purposefully clean heavy spot parts and stain areas, save water resources, improve cleaning quality, effectively remove stains, flexibly provide high-pressure water guns or force control floating brush heads, realize different cleaning modes of robots in various scenes, design various cleaning paths for different types of equipment or parts to be cleaned, independently package each cleaning path into a cleaning path process package, and improve the operability of the system. Aiming at the scheme, the technical problems that 1, the current technology mainly determines the cutting path of the robot to the target object through manual teaching or off-line programming, has large programming workload and long time consumption for complex-shape and frequently-replaced cutting tasks, has higher requirements on professional skills and experience of operators, and the robot can usually cut according to preset programs and paths, so that the cutting strategy and paths are difficult to automatically adjust in real time like according to actual cutting conditions, which results in insufficient intellectualization of the current path planning method and has to be improved in processing precision. 2. The prior art lacks analysis of the processing effect of the processed target object, further lacks scheme planning of secondary processing of the outer surface of the object, and the neglect of the aspect of the prior art can lead to the problems of unstable processing quality, reduced processing precision and the like of the object, further causes the prior art to lack perfection and rationality. Disclosure of Invention The application aims to provide an intelligent path planning method for robot operation outside an object, which solves the problems in the background technology. In order to solve the technical problems, the application provides an intelligent path planning method for robot operation outside an object, which comprises the following processing path analysis of the outer surface of the target object. And step two, evaluating the processing effect of the robot on the target object through the primary following processing path, and judging whether the target object needs secondary processing or not. And thirdly, analyzing a secondary processing scheme, namely analyzing the primary processing effect of the target object to obtain the secondary processing scheme. S1, carrying out region segmentation on the outer surface of the target object based on a normal vector of the outer surface of the target object, and acquiring vertical visual angle depth images corresponding to the regions. S2, taking the vertical visual angle depth image of each region as input, and outputting the optimal slice direction and the optimal slice width of each region through a neural network technology. And S3, carrying out point cloud slicing on each region based on the optimal slicing direction and the optimal slicing width to obtain the contour based on the point cloud slicing of each region, generating a slice follow-up path of each region based on the contour and the normal vector of the point cloud slicing of each region, and analyzing the slice follow-up path sequence of each region to obtain the primary follow-up processing path of the outer surface of the target object. Preferably, the method comprises dividing the region of the outer surface of the target object based on the normal vector of the outer surface of the target object, and acquiring the