insertPointCloud

Insert point cloud data into 3-D signed distance map

Since R2024b

Syntax

insertPointCloud(sdm3D,origin,points)

Description

insertPointCloud(sdm3D,origin,points) inserts a point cloud into the 3-D signed distance map by casting rays from the specified origin to each of the specified points, and updates the distance values of all voxels the digital differential analyzer (DDA) algorithm determines each ray hits.. The FullTracing property of the 3-D signed distance map determines the raycasting type. See FullTracing for more information.

example

Examples

collapse all

Build Signed Distance Map from Point Cloud Data

Open Live Script

Read and plot the triangulation data from the L-membrane STL. You can use this to compare with the signed distance map.

triL = stlread("L-Membrane.stl");
trisurf(triL)
title("L-Membrane STL")
[az,el] = view;
axis equal

Create an empty 3-D signed distance field (SDF) and load the point cloud data for the L-membrane from a MAT file. The MAT file contains point cloud data and the sensor origin associated with each point cloud data.

sdm3D = signedDistanceMap3D(Resolution=50);
load LMembranePC.mat

Insert each point cloud into the signed distance field and show the updated 3-D SDF.

for i = 1:size(ptcloud,1)
    insertPointCloud(sdm3D,origin(i,:),ptcloud{i});
    show(sdm3D,Colorbar="on");
    view(az,el)
    axis equal
    drawnow
    pause(.25)
end
title(["3-D SDF of L-Membrane"])

Get all active voxels in the 3-D SDF.

vox = activeVoxels(sdm3D)

vox = struct with fields:
           ID: 1
      Centers: [49889x3 double]
    Distances: [49889x1 double]
        Sizes: [49889x1 double]

For demonstrative purposes, use a random xyz-offset from the voxel centers of the first three voxels as query points. Then get the distance and gradient using those query points.

querypts = vox.Centers(1:3,:) + 0.1*rand(3,3);
d = distance(sdm3D,querypts)

g = gradient(sdm3D,querypts)

g = 3×3

    0.1924    1.5773   -1.4002
    0.1689    1.6524   -0.8267
   -0.0715   -0.2335    0.3677

Generate a mesh from the 3-D signed distance field.

[vertices,faces] = mesh(sdm3D)

vertices = 18860×3

   -0.0100    0.0068    0.1500
   -0.0300    0.0059    0.1500
   -0.0100    0.0100    0.1610
   -0.0300    0.0100    0.1565
    0.0100    0.0047    0.1500
   -0.0100   -0.0100    0.1409
    0.0100   -0.0100    0.1421
    0.0100    0.0025    0.1700
   -0.0010    0.0100    0.1700
    0.0100    0.0094    0.1900
      ⋮

faces = 37094×3

     2     1     3
     4     2     3
     7     1     6
     5     1     7
     9     3     1
     9     1     8
     8     1     5
     8    10    11
     9     8    11
    10    12    11
      ⋮

Visualize the mesh data.

meshTri = triangulation(faces,vertices);
trisurf(meshTri)
axis equal
title("Mesh from 3-D SDF")

If needed, you can use this mesh data to create a collision mesh using V-HACD. See collisionVHACD (Robotics System Toolbox) for more information.

Input Arguments

collapse all

`sdm3D` — 3-D signed distance map
`signedDistanceMap3D` object

3-D signed distance map, specified as a signedDistanceMap3D object.

`origin` — Origin of point cloud data
three-element vector

Origin of the point cloud data, specified as a three-element vector representing an xyz-point in the world frame. Units are in meters.

Example: [0 0 0]

`points` — Point cloud data
N-by-3 matrix

Point cloud data, specified as an N-by-3 matrix. Each row represents an xyz-point observed from a sensor at origin. N is the total number of points in the point cloud.

If the points are measured with respect to the sensor frame, you must transform the points to the world frame.

Extended Capabilities

C/C++ Code Generation
Generate C and C++ code using MATLAB® Coder™.

Version History

Introduced in R2024b