Global Optimality of Pose Graph Optimization

Key points

• In this paper, we investigate the optimization landscape and establish a sufficient condition of global optimality for PGO.1
• The ultimate goal of this paper is to theoretically confirm that the optimization landscape is benign and attractive around the global minimizer, if noise is smaller than an explicit and a priori bound.1
• First, we identify a proper basin of attraction (BOA) containing global minimizer, and characterize its geometric property by showing that its radius is controlled by noise in an explicit way.1
• Then, by exploring the first order optimality conditions of PGO, we prove that a local minimizer is globally optimal if a noise-controlled criterion is satisfied.1

Sources (1)

• [1]Global Optimality of Pose Graph Optimization
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 10, 07:12 AM

  Pose Graph Optimization (PGO) is a classical and widely applied problem in state estimation for robotics, playing a fundamental role in localization, mapping, and navigation.

  In this paper, we investigate the optimization landscape and establish a sufficient condition of global optimality for PGO.

Extractive summary: sentences quoted from the sources.