METHOD AND DEVICE FOR PROCESSING A 3D POINT CLOUD REPRESENTING SURROUNDINGS

Abstract

A method and to a device for processing a 3D point cloud representing surroundings, which is generated by a sensor. Initially, starting cells are identified based on ascertained starting ground points within the 3D point cloud which meet at least one predefined ground point criterion with respect to a reference plane divided into cells. Thereafter, cell planes are ascertained for the respective starting cells of the reference plane. Thereafter, estimated cell planes and ground points are ascertained for candidate cells deviating from the starting cells based on the cell planes of the starting cells, which are subsequently converted into final cell planes. As a result of such a cell growth originating from the starting cells, the cells of the reference plane are iteratively run through and processed so that the 3D point cloud is reliably classifiable into ground points and object points based on this method.

Claims

1-12. (canceled)

13. A method for processing a 3D point cloud representing surroundings, comprising the following steps: receiving a 3D point cloud ascertained based on a sensor; ascertaining starting ground points within the 3D point cloud, those points of the 3D point cloud being classified as the starting ground points which are situated within a first predefined distance from the sensor and meet at least one predefined ground point criterion with respect to a reference plane, the reference plane being a plane which is predefined with respect to the sensor and, being a plane which represents a ground surface; dividing at least one sub-area of the reference plane into a plurality of cells, and ascertaining points of the 3D point cloud which correspond to each of the cells, points corresponding to a respective cell being those points which are enclosed by the respective cell in a perpendicular projection onto the reference plane; establishing those cells as starting cells which have a first predefined minimum number of corresponding starting ground points including at least three starting ground points; ascertaining a cell plane for each respective starting cell of the starting cells, the cell plane being ascertained in such a way that it approximates a position of the starting ground points of the respective starting cell according to a predefined calculation rule; ascertaining candidate cells within the reference plane, each candidate cell: being a cell for which no cell plane has been ascertained yet, abutting at least one further cell for which a cell plane has already been ascertained, and including a second predefined minimum number of corresponding points of the 3D point cloud; calculating an estimated cell plane for each respective candidate cell from all cell planes which are present in cells which directly abut the respective candidate cell; ascertaining cell plane candidate points for each candidate cell, each point of the 3D point cloud being ascertained as a cell plane candidate point when: the point corresponds to the candidate cell, and its smallest distance from the estimated cell plane of the candidate cell and/or from the reference plane does not exceed a second predefined distance; and ascertaining a cell plane for each respective candidate cell, each cell plane being ascertained in such a way that it approximates a position of the cell plane candidate points of the respective candidate cell according to the predefined calculation rule.

14. The method as recited in claim 13, wherein: (i) the cells which divide the reference plane abut one another without interruption, and/or are triangular or quadrangular cells, and/or (ii) a shape and/or an extension of the respective cells is adapted as a function of a maximum resolution of the sensor, and/or existing boundary conditions.

15. The method as recited in claim 13, wherein the at least one predefined ground point criterion is met when: a shortest distance between respective points of the 3D point cloud and the reference plane does not exceed a third predefined distance, and/or all vector products which result from possible combinations of vectors between a respective point of the 3D point cloud to be considered and at least two points directly adjoining the respective point have a maximum permissible first angular deviation with respect to a normal of the reference plane, and/or an average vector, ascertained from the vector products for each point of the 3D point cloud, has a maximum permissible second angular deviation from the normal of the reference plane.

16. The method as recited in claim 13, wherein those points of the 3D point cloud are classified as object points: which are situated within the first predefined distance from the sensor, whose smallest distance from the reference plane or from the estimated cell plane or from the cell plane exceeds a predefined object minimum distance, and whose vector products with directly adjoining points exceed a third predefined angular deviation with respect to a normal of the reference plane.

17. The method as recited in claim 13, wherein a subset or all points of the 3D point cloud, regardless of a prior classification as ground points, is classified as object points when: their closest adjoining point was classified as an object point, a shortest distance from the closest adjoining point does not exceed a fourth predefined distance, and a distance between the point and the closest adjoining point with respect to the reference plane does not exceed a fifth predefined distance.

18. The method as recited in claim 13, wherein the calculation rule effectuates a minimization of a sum of squared deviations between the cell plane and the corresponding starting ground points and/or cell plane candidate points.

19. The method as recited in claim 13, wherein each cell plane is only ascertained for a respective cell or used in a downstream processing step when: a value for a distribution of starting ground points or cell plane candidate points within a respective cell reaches a predefined minimum distribution value, the value for the distribution being greater the further the points of a cell which in each case have the largest distance from a center of the cell are situated away from the center, and the more uniformly the points are distributed within the cell, and/or the cell plane does not exceed a maximum permissible angle of inclination with respect to the reference plane.

20. The method as recited in claim 13, wherein initially an individual weighting of the cell planes of adjoining cells takes place during the calculation of the estimated cell plane, the individual weighting being higher: the higher a value is for the distribution of starting ground points and/or cell plane candidate points in the respective adjoining cell, and/or the lower a sum is of squared deviations between the cell plane and corresponding starting ground points and/or cell plane candidate points in the respective adjoining cell.

21. The method as recited in claim 13, wherein all points of the 3D point cloud within a cell including an ascertained cell plane which were not yet classified as ground points or as object points are classified as ground points when their smallest distance from the cell plane of the cell does not exceed the second predefined distance.

22. The method as recited in claim 13, wherein a surroundings recognition is carried out in a surroundings recognition system based on the ascertained cell planes and/or the ascertained ground points and/or the ascertained object points.

23. The method as recited in claim 13, wherein the sensor is a LIDAR sensor of a transportation device.

24. A device for processing a 3D point cloud representing surroundings, the device comprising: an evaluation unit configured to: receive a 3D point cloud ascertained based on a sensor; ascertain starting ground points within the 3D point cloud, those points of the 3D point cloud being classified as the starting ground points which are situated within a first predefined distance from the sensor and meet at least one predefined ground point criterion with respect to a reference plane, the reference plane being a plane which is predefined with respect to the sensor and, being a plane which represents a ground surface; divide at least one sub-area of the reference plane into a plurality of cells, and ascertaining points of the 3D point cloud which correspond to each of the cells, points corresponding to a respective cell being those points which are enclosed by the respective cell in a perpendicular projection onto the reference plane; establish those cells as starting cells which have a first predefined minimum number of corresponding starting ground points including at least thee starting ground points; ascertain a cell plane for each respective starting cell of the starting cells, the cell plane being ascertained in such a way that it approximates a position of the starting ground points of the respective starting cell according to a predefined calculation rule; ascertain candidate cells within the reference plane, each candidate cell: being a cell for which no cell plane has been ascertained yet, abutting at least one further cell for which a cell plane has already been ascertained, and including a second predefined minimum number of corresponding points of the 3D point cloud; calculate an estimated cell plane for each respective candidate cell from all cell planes which are present in cells which directly abut the respective candidate cell; ascertain cell plane candidate points for each candidate cell, each point of the 3D point cloud being ascertained as a cell plane candidate point when: the point corresponds to the candidate cell, and its smallest distance from the estimated cell plane of the candidate cell and/or from the reference plane does not exceed a second predefined distance; and ascertain a cell plane for each respective candidate cell, each cell plane being ascertained in such a way that it approximates a position of the cell plane candidate points of the respective candidate cell according to the predefined calculation rule.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Exemplary embodiments of the present invention are described hereafter in greater detail with reference to the figures.

[0034] FIG. 1 shows one example of a 3D point cloud processed based on a method according to an example embodiment of the present invention.

[0035] FIG. 2 shows one example of a candidate cell and points of the 3D point cloud corresponding to the candidate cell.

[0036] FIG. 3 shows one example of an iterative cell growth according to an example embodiment of the present invention along a ground surface.

DETAILED DESCRIPTION OF EXAMPLE EMBODIMENTS

[0037] FIG. 1 shows one example of a 3D point cloud 20 which is processed based on a method according to the present invention and detected with the aid of a LIDAR sensor 10 of a road vehicle. The points of 3D point cloud 20 are projected onto a reference plane 40 whose orientation essentially corresponds to an orientation of a roadway surface in the immediate surroundings of LIDAR sensor 10. The representation in FIG. 1 corresponds to a top view onto reference plane 40. LIDAR sensor 10 is connected, in terms of information technology, to an evaluation unit 92, in this case a microcontroller, of a control unit 90 of the road vehicle. Evaluation unit 92 is configured to receive the 3D point cloud detected by LIDAR sensor 10 and to process it according to the method according to the present invention. A result of the processing by evaluation unit 92 is transferred, for example, to a surroundings recognition system, which may also be an integral component of control unit 90.

[0038] It shall generally be pointed out that, for reasons of a simplified representation, only individual elements are representatively denoted by reference numerals in FIG. 1. These reference numerals implicitly also apply accordingly for identically represented further elements in FIG. 1.

[0039] With the aid of evaluation unit 92, 3D point cloud 20 is initially divided into a plurality of square cells 50 abutting one another without interruption. Thereafter, those points of 3D point cloud 20 are considered which are situated within a first predefined distance 30 of 15 m from LIDAR sensor 10. For all points within first distance 30, it is checked in each case whether a shortest distance of these points from reference plane 40 does not exceed a third predefined distance of 0.3 m. In the event that the third distance is not exceeded for a respective point, the respective point is established as a starting ground point 22. All ground points 24 established as starting ground points 22 are marked here by a single ring which surrounds starting ground points 22.

[0040] Moreover, points of 3D point cloud 10 are identified with the aid of evaluation unit 92 as obvious object points 26 (i.e., points which were obviously caused by an object in the surroundings) which are situated within first distance 30 and whose smallest distance from reference plane 40 is above a predefined object minimum distance of 0.3 m and whose vector products 72 with directly adjoining points in each case exceed a third predefined angular deviation of 45° with respect to a normal 45 of reference plane 40. These obvious object points 26 are in each case identified by two concentric rings which surround the obvious object points 26.

[0041] Thereafter, those cells 50 within first distance 30 which have a required minimum number of corresponding starting ground points 22, the minimum number corresponding to a value of 5 here, are established as starting cells 52.

[0042] Thereafter, a cell plane 54 is ascertained for each starting cell 52, cell plane 54 approximating the position of starting ground points 22 in the respective cell 50 in such a way that a sum of the squared deviations of the shortest distances of starting ground points 22 of the cell from cell plane 54 is minimized.

[0043] Thereupon, candidate cells 60 abutting starting cells 52 are ascertained, which must each include a minimum number of two points here to be considered candidate cells 60. For each candidate cell 60, an estimated cell plane 62 is now ascertained from the previously ascertained cell planes 54 of adjoining starting cells 52 in that an average plane is ascertained from all directly abutting, adjoining cell planes 54.

[0044] In each candidate cell 60, cell plane candidate points 64 are now ascertained whose smallest distance from estimated cell plane 62 does not exceed a second predefined distance of 0.3 m.

[0045] Thereafter, a cell plane 54 is ascertained for each candidate cell 60, cell plane 54 approximating the position of cell plane candidate points 64 in the respective candidate cell 60 in such a way that a sum of the squared deviations of the shortest distances of cell plane candidate points 64 of candidate cell 60 from cell plane 54 is minimized.

[0046] Proceeding from the previously processed candidate cells 60, candidate cells 60 which abut these candidate cells 60 and which have not yet been processed are identified according to the above description, and above-described method steps are employed to also ascertain cell planes 54 and ground points 24 corresponding thereto for these candidate cells 60. This is continued until all cells 50 of reference plane 40 have been processed.

[0047] In the event that the boundary conditions for ascertaining cell planes 54 and ground points 24 for the respective cells 50 (starting cells 52 and candidate cells 60) should not be met, an estimated cell plane 62 is used for these cells 50, which is ascertained from all existing cell planes 54 in adjoining cells 50 (starting cells 52 and candidate cells 60).

[0048] Thereafter, those points of 3D point cloud 20 are classified as object points 26 which are situated within the first predefined distance 30 from sensor 10, whose smallest distance from cell plane 54 exceeds a predefined object minimum distance, and whose vector products 72 with directly adjoining points exceed a third predefined angular deviation of 45° with respect to a normal of cell plane 54.

[0049] FIG. 2 shows one example of a candidate cell 60 and points of 3D point cloud 20 corresponding to candidate cell 60. Four points of 3D point cloud 20 are shown, which in a projection onto a reference plane 40 (not shown), which represents a ground surface and which is divided into a plurality of mutually abutting cells 50, are situated within cell 50 of reference plane 40 which corresponds to candidate cell 60. An estimated cell plane 62 of candidate cell 60 was ascertained based on cell planes 54 which exist in cells 50 which abut candidate cell 60. Now those points of the 3D point cloud whose distance from estimated cell plane 62 does not exceed a second predefined distance of 0.2 m are identified as cell plane candidate points 64. This applies both to points above and below the estimated cell plane 62. Three of these points meet this distance criterion here, while one of the points does not meet this criterion and accordingly is rated as a non-classified point 28. Based on the points identified as cell plane candidate points 64, thereafter a final, i.e. non-estimated, cell plane 54 is ascertained for this candidate cell 60.

[0050] FIG. 3 shows one example of an iterative cell growth according to the present invention along a ground surface. FIG. 3 may be considered a detail from FIG. 1, further cells 50 of reference plane 40 already having been processed here compared to FIG. 1. To avoid repetition, only differences compared to FIG. 1 are therefore described hereafter. Shown is a starting cell 50 and four candidate cells 60, for each of which a cell plane 54 was ascertained. For each of these cells 50, 60, a local normal vector 74 was ascertained in each case, which is an average vector which was ascertained based on all vector products 72 between ground points within a cell 50, 60. Local normal vectors 74 are subsequently used to ascertain an angular deviation between the respective local normal vectors 74 and a normal vector 45 of reference plane 40. If the ascertained angular deviation is greater than a predefined second angular deviation of 15°, the respective cell plane 50, 60 is not taken into consideration in a downstream processing since it is then not regarded as a reliable ground cell. Moreover, an estimated cell plane 62 of a candidate cell 60 to be instantaneously processed is shown in FIG. 3.