Analysis for Graph-Based SLAM Algorithms
Simultaneous Localization and Mapping (SLAM) is the computational problem of constructing or updating a map of an unknown environment while simultaneously keeping track of an agent’s location within it. The back-end of SLAM can be classified into two processing methods: filter methods (e.x. Kalman Filter), and nonlinear optimization (e.x. graph-based optimization). Although filter methods used to be popular among researchers in SLAM, state-of-the-art research topics in the SLAM field are mostly in graph-based optimization recently. This is because filter methods, such as Kalman Filter, are not suitable for large environments with a huge amount of data in visual SLAM. In visual SLAM, the number of camera poses and image feature points is huge, but the Jacobian in graph-based method is sparse, which means it can speed up. Compared to filter methods, the efficiency of nonlinear optimization is significantly higher in visual SLAM. That’s why our project focus will be on graph-based optimization.