GraphSLAM

Experiment in the fourth grade of undergraduate

Graph SLAM algorithm

Graph SLAM is graph-based Simultaneous Localization and Mapping.

Requirements (library dependency)

You need Python 3.6 or later to run GraphSLAM.

• numpy
• scipy
• matplotlib
• pandas

Datasets

You can make sample datasets locally by MakeSmaple.py. Like this:

mkdir data
cd src
python MakeSample.py

You will see sample0 directory in data.

You can also download datasets online here.

Quick start

First, install the libraries and change the current directory to src.

cd src
python main.py ../data/sample0 0

The last argument is index of methods for inverse matrix calculation. You can choose the number between 0 to 2.

• "0" : numpy package as dense
• "1" : sum of block diagonal as dense
• "2" : scipy package as sparse

References

[1] Sebastian Thrum, Wolfram Burgard, and Dieter Fox, 上田隆一(訳)(2017)「確率ロボティクス」