A Spanning Tree-Based Multi-Resolution Approach for Pose-Graph Optimization

This paper proposes a computationally efficient method for pose-graph optimization that makes use of a multi-resolution representation of pose-graph transformation constructed on a spanning tree. It is shown that the proposed spanning tree-based hierarchy has a number of advantages over the previously known serial chain-based hierarchy in terms of preservation of sparsity and compatibility with parallel computation. It is demonstrated in numerical experiments using several public datasets that the proposed method outperforms a state-of-the-art solver for large-scale datasets.