SLAM++ is an implementation of incremental nonlinear least squares solvers, containing lighting fast linear algebra on sparse block matrix operations. It is aimed for the use in 3D reconstruction and SLAM online applications. The code has been exhaustively tested and compared with the existing similar implementations and constitutes the base for more than ten high impact publications.
Comparing to other similar software (g2o, gtsam, isam), SLAM++ is based on an efficient sparse, block matrix data-structure that facilitates numerical and structural changes in the system matrix as well as speeds up matrix arithmetic operations and matrix factorization. This allowed for very efficient solutions to incremental problems, where the size of the state increases every step as well as for efficient covariance recovery. A comprehensive explanation of the block-matrix manipulation to obtain the covariances is given at:
Australian Centre for Robotic Vision
2 George Street Brisbane, 4001
+61 7 3138 7549