Sparsify pose graph and maintain the uncertainty. (Only supported on Linux)
This repository is based on several research papers
-
Nonlinear factor recovery for long-term SLAM. This paper proposed the NFR and re-implemented GLC method to sparsify pose graph. Both two method implementations have been included.
-
Generic Node Removal for Factor-Graph SLAM. This paper proposed the GLC method to sparsify pose graph. It addresses the issue that the direct composition between two nodes is not equivalent to marginalization (i.e. nodes delete).
-
Estimating uncertain spatial relationships in robotics. This paper gives an intuitive explanation for preliminary math.
You can check the test_xx.cpp in 'src/' about the algorithm details.
Refer to the CMakelists.txt for details