Search

EP-3989164-B1 - METHOD FOR IDENTIFYING A DISCONTINUITY ON THE SURFACE OF AN OBJECT

EP3989164B1EP 3989164 B1EP3989164 B1EP 3989164B1EP-3989164-B1

Inventors

  • CASTANO, MARCO

Dates

Publication Date
20260506
Application Date
20211026

Claims (13)

  1. A computer-implemented method for identifying a discontinuity on a surface of an object placed within a work environment, said method including: a) digitizing said work environment to obtain a 3D point cloud; b) determining, within said work environment, a search volume including said discontinuity and extracting from said point cloud the points belonging to said search volume; c) considering the points of said point cloud within said search volume, determining the normals to the surfaces defined by said points; d) determining, for each of these normals, its orientation; and e) analyzing the orientation of said normals in said search volume to identify said discontinuity in the presence of an anomalous variation in the orientation of the normals, said step of analyzing the orientation of said normals comprises, for the points of said point cloud inside said search volume, calculating the angles between the normals and determining that two normals are similar if the angle between them is less than a predefined threshold, identifying at least two groups of homogeneous points associated with normals similar to each other and determining the presence of said anomalous variation at intermediate points between said at least two groups of homogeneous points.
  2. The method according to claim 1, said method comprising filtering the points of said 3D point cloud on the basis of the Euclidean distance between them, said filtering comprising calculating, for each point, an average distance with respect to a number, N, of adjacent points and removing from said 3D point cloud the points that have an average distance from their N adjacent points greater than the value of a filtering threshold.
  3. The method according to claim 2, wherein said filtering further comprises calculating a standard deviation of the average distances calculated for each point of said 3D point cloud and defining, for each point of said 3D point cloud, the value of said filtering threshold, f_th, in the following manner: f_th = m + stdv * sX , where m is said average distance, stdv is said standard deviation and sX is a predefined factor.
  4. The method according to any one of the preceding claims, said method comprising localizing one or more known reference objects or edges in the 3D point cloud and repositioning said search volume based on said localization of one or more known reference objects or edges.
  5. The method according to any one of the preceding claims, wherein said step of determining the normals to the surfaces defined by the points comprises performing a mathematical triangulation of the points included in said search volume to determine a set of triangles such that the union of the surfaces of said triangles represents the surface of the objects included in said search volume.
  6. The method according to claim 5, wherein said step of determining the normals comprises calculating said normals with respect to the surfaces of said triangles and said step of analyzing the orientation of said normals comprises calculating an orientation of a versor of the normal of each triangle with respect to a reference versor as the component of the versor of the normal in a direction defined by the reference versor.
  7. The method according to claim 6, wherein said calculating an orientation of a versor of the normal of each triangle comprises calculating the cosine of the angle, T, between said versor of the normal and said reference versor.
  8. The method according to claim 6 or 7, wherein said step of analyzing the orientation of said normals comprises checking that all the normals have said component with the same sign and, if one of said normals has said component with an opposite sign, inverting the sign of said component with an opposite sign.
  9. The method according to any one of the preceding claims, wherein said step of analyzing the orientation of said normals comprises selecting groups of homogeneous points comprising a number of points between a predefined minimum number and a predefined maximum number.
  10. The method according to claim 9, wherein said step of analyzing the orientation of said normals comprises building a depth map starting from the selected groups of homogeneous points and identifying said discontinuity on said depth map at a hole on said map.
  11. A system for identifying a discontinuity on a surface of an object placed within a work environment, said system including: - an acquisition unit (11) configured to digitize said work environment to obtain a 3D point cloud; and - a processing unit (12) configured to: - extract from said point cloud the points belonging to a search volume comprising said discontinuity; - considering the points of said point cloud within said search volume, determine the normals to the surfaces defined by said points; - determine, for each of said normals, its orientation; and - analyze the orientation of said normals in said search volume to identify said discontinuity in the presence of an anomalous variation in the orientation of the normals, wherein the processing unit (12) is further configured to, for the points of said point cloud inside said search volume, calculate the angles between the normals and determining that two normals are similar if the angle between them is less than a predefined threshold, identify at least two groups of homogeneous points associated with normals similar to each other and determine the presence of said anomalous variation at intermediate points between said at least two groups of homogeneous points.
  12. The system according to claim 11, wherein said acquisition unit (11) comprises a digitizing sensor which implements an artificial vision system and comprises a stereoscope system which uses two cameras and an LED projector.
  13. The system according to claim 11 or 12, said system further comprising an actuation unit (13) for sending commands to a robot (3) to carry out a work on said object at the identified discontinuity.

Description

TECHNICAL FIELD The present invention relates in general to the field of automated systems for controlling the surfaces of objects in an industrial environment. In particular the present invention relates to a method for identifying and locating a discontinuity (e.g., a weld, a joint, etc.). on the surface of an object, for example a pipe. PRIOR ART In the industrial field, it is known to use methods and systems for inspecting the surface of an object, such as a pipe, in order to check for the intact condition of the part and if necessary guide an anthropomorphic robot towards a possible discontinuity on this surface in order to carry out work in that area (for example, a repair). Possible discontinuities on the surface of a pipe are associated with, for example, welds or joints or junctions or with defects, irregularities or anomalies present on the surface of the said pipe. Hitherto, to inspect the surface of an object in an industrial environment, in particular a pipe, manual methods carried out by human operators (for example, visual inspections) or automated systems are used. The automated systems involve the use of robots and laser triangulation vision systems and/or 2D vision systems. These systems are able to identify a discontinuity on the surface of a pipe by means of analysis of the diameter of the said pipe (for example, calculating the difference in diameter of the pipe at the point of discontinuity) and/or indirectly by means of calculation of a fixed distance with respect to reference objects which are easier to recognize. Wang Xiangfei et al., "Point cloud 3D parent surface reconstruction and weld seam feature extraction for robotic grinding path planning", The International Journal of Advanced Manufacturing Technology, Springer, London, vol. 107, No. 1-2, 20 February 2020, pages 827-841, describes a method for planning the path of a robot to automatically remove welds from a surface. CN 111 192 225 B describes a method for determining whether defects are present or not on piping. SUMMARY OF THE INVENTION The inventor has noted that the vision systems currently used to check for discontinuities on pipes have a number of drawbacks. In particular, the inventor has noted that when the difference in diameter between the two pipe portions joined together at, for example, a joint is not significant and/or when the position of the discontinuity, with respect to known reference objects, is not repetitive, it becomes difficult to be able to precisely determine the position of these discontinuities. Consequently, it also becomes difficult to be able to determine the trajectory which the anthropomorphic robot must follow to safely reach the area of the discontinuity which is being monitored and analyzed. The Applicant has therefore defined the object of providing a method and a system which is able to identify (i.e., recognize and locate) a discontinuity on the surface of an object, in particular a pipe, in a more precise and reliable manner compared to the known methods. The aforementioned object, in addition to other objects, is achieved by a method and a system for identifying a discontinuity on the surface of an object, in particular a pipe, which involves the use of a 3D vision system able to reconstruct the scene located opposite the system itself in the form of a point cloud, and an algorithm able to analyze the aforementioned point cloud in order to recognize and determine the position of a possible discontinuity. In the following present description and in the claims, the term "discontinuity" will be used to indicate an area or zone of a surface of an object comprising a variation in the thickness of this surface at a position where, for example, a weld is located. In case the object is a pipe or two interconnected pipes, a discontinuity may occur at the location of a junction or joint between pipes or pipe portions, or of a weld, where there is a difference in diameter between the pipes or the pipe portions. According to a first aspect, the present invention provides a method for identifying a discontinuity on a surface of an object placed within a work environment as defined in independent claim 1. Preferably, the method comprises filtering the points of the 3D point cloud on the basis of the Euclidean distance between them, said filtering comprising calculating, for each point, an average distance with respect to a number, N, of adjacent points and removing from said 3D point cloud the points which have an average distance from their N adjacent points greater than the value of a filtering threshold. Preferably, the filtering operation further comprises calculating a standard deviation of the average distances calculated for each point of the 3D point cloud and defining, for each point of the 3D point cloud, the value of the filtering threshold, f_th, in the following manner:f_th=m+stdv*sX, where m is the average distance, stdv is the standard deviation and sX is a predefined factor (for exa