muchensun/range_mi

Functions to compute expected information gain in robotic exploration.

Range Mutual Information

This package defines functions that compute the amount of information that is expected to be gained about an incomplete occupancy map from a potential range measurement. Range measurement sources could include lidar, sonar, depth cameras, etc. Below is an example output produced by this code. The package can be built standalone or as a ROS library.

Usage

C++ API

Clone the library and build with cmake:

git clone https://github.com/sportdeath/range_mi.git
cd range_mi
mkdir build
cd build
cmake ..
make

To build the tests you can add a flag to cmake before making:

cmake -DBUILD_TESTS=ON ..
make

To run a test, for example the test/demo.cpp test, execute:

./test/demo

The other test cases compare the output of the algorithm to numerical integration and benchmark speed.

Python API

After building the C++ library as described above also install the library:

sudo make install

Then compile the python library (requires cython):

python setup.py build_ext --inplace

This builds a function grid_mi that simply takes as input a 2D occupancy map and returns a 2D mutual information surface (both numpy arrays). A simple demo is included in the pysrc/demo.py file.

ROS Demo Nodes

Clone the library into your catkin workspace:

cd ~/catkin_ws/src
git clone https://github.com/sportdeath/range_mi.git
cd ~/catkin_ws
catkin_make

Then to compute the mutual information of an occupancy grid, you can run:

roslaunch range_mi occupancy_grid_mi.launch

In rviz the occupancy grid map will be displayed on the topic /map and the mutual information surface will be displayed on the topic /mi_map. The topics names, number of range measurement beams and other parameters are available in the launch/params.yaml file. By running the node/mi_grid_heatmap.py script, the mutual information surface can be saved as colored image like the one above.

By using the "Publish Point" tooltip in rviz, you can click on a point in the map to condition the mutual information on a measurement being made at that point. This will likely reduce the information gain within the view of the clicked point.

We have also included another script that computes the mutual information by averaging the information gain over random occupancy assigments. This will take much longer than the included algorithm, but it will eventually converge to the same result. Run it on a map as follows:

roslaunch range_mi monte_carlo_mi.launch