A survey of back-end optimization method for graph-based SLAM under large-scale environment

Abstract:

Graph optimization-based SLAM is the main method under large-scale environment. The framework of this method is composed of two parts, front-end and back-end. Be a continuation paper of our previous one, the four main back-end optimization approaches, which include least square, stochastic gradient descent, relaxation, manifold optimization, and the correspondent literatures are introduced, and two map evaluation methods are presented, that is  χ² error based and MSE error based. The trends of graph optimization-based SLAM method are predicted.