This new “rotation averaging” method, an important step in 3D reconstruction and robot mapping, is described in our ECCV 2020 paper, https://arxiv.org/abs/2008.02737,
Shonan Rotation Averaging: Global Optimality by Surfing SO(p)^n
Frank Dellaert(1), David M. Rosen(2), Jing Wu(1), Robert Mahony(3), and Luca Carlone(2)
- Georgia Institute of Technology, Atlanta, GA
- Massachusetts Inst. of Technology, Cambridge, MA
- Australian National University, Canberra, Australia
Abstract: Shonan Rotation Averaging is a fast, simple, and elegant rotation averaging algorithm that is guaranteed to recover globally optimal solutions under mild assumptions on the measurement noise. Our method employs semidefinite relaxation in order to recover provably globally optimal solutions of the rotation averaging problem. In contrast to prior work, we show how to solve large-scale instances of these relaxations us- ing manifold minimization on (only slightly) higher-dimensional rotation manifolds, re-using existing high-performance (but local) structure-from- motion pipelines. Our method thus preserves the speed and scalability of current SFM methods, while recovering globally optimal solutions.