Python¶

Overview¶

The gmm_occupancy_modeling_examples package contains python scripts for sampling from GMMs and updating an occupancy grid map for an entire dataset of a mine.

Initialize the Python Environment¶

To initialize the python paths properly, you should source workon at the top level sandbox directory.

cd gira3d-occupancy-modeling
source workon

Download Data¶

The data directory contains a script that you will need to execute in order to download all of the data associated with this tutorial.

pip install gdown
cd gira3d-occupancy-modeling/data
./download.sh

The complete dataset is < 200MB so it should download fairly quickly.

Running the Example¶

Go to the gmm_occupancy_modeling_examples/python directory under wet/src and run the occupancy_modeling_example.py file.

cd wet/src/gmm_occupancy_modeling_examples/python
python occupancy_modeling_example.py

Detailed Explanation of the Script¶

First, we initialize the Grid3D data structure.

p = Parameters()
with open('../config/grid3d.yaml') as file:
    params = yaml.safe_load(file)
    p = convert_dict_to_params(params)
TRIMMED_MAX_RANGE = MAX_RANGE - p.resolution
OCCUPANCY_THRESH = p.occupancy_threshold

grid = Grid3D(p)

Next, the GMMs are load from file and points are sampled.

gmmfile = GMM_DIR + str(i) + '.gmm'
gmm = load_gmm_from_file(gmmfile)
ret = gmm.sample(2e4)
pts = ret[0]

The points are transformed into the world frame using the ground truth data stored in odometry.pkl.

R = Twc[0:3, 0:3]
t = np.reshape(Twc[0:3, 3], (3,1))
pts = np.matmul(R, np.transpose(pts)) + np.matlib.repmat(t, 1, np.shape(pts)[0])

Each point is raytraced through the occupancy grid map

for j in range(0, np.shape(pts)[0]):
    st = Point(t[0,0], t[1,0], t[2,0])
    en = Point(pts[j,0], pts[j,1], pts[j,2])
    grid.add_ray(st, en, TRIMMED_MAX_RANGE)

Finally, the occupied points are recovered from the occupancy grid map and visualized.

n = p.width*p.height*p.depth
all_pts = None
for i in range(0,p.width*p.height*p.depth):
    if grid.get(i).logodds > OCCUPANCY_THRESH:
        pt =  grid.get_point(i)
        if all_pts is None:
            all_pts = np.array([pt.x, pt.y, pt.z])
        else:
            all_pts = np.vstack((all_pts, np.array([pt.x, pt.y, pt.z])))

viz = Open3DVisualizer()
viz.plot3d(all_pts)

The expected output is: Plotting occupied points which plots the points that exceed the occupancy threshold from the occupancy grid map.

Creating GMMs using Sklearn¶

A script is provided to learn GMMs using sklearn. In order to do so, you will need to install scikit-learn.

pip install scikit-learn

The script is provided in the gmm_occupancy_modeling_example/python/utils/create_and_save_gmm_example.py script.

python create_and_save_gmm_example.py

The GMM will be saved in your current working directory.

Operating Systems¶

These tutorials have been tested on the following operating systems:

Ubuntu 20.04
Ubuntu 18.04