Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

IPC: Incremental Probabilistic Consensus-based Consistent Set Maximization for SLAM Backends

Emilio Olivastri, Alberto Pretto

TL;DR

IPC addresses robustness in SLAM back-ends by incrementally building a maximally consistent set of loop closures using a consensus mechanism. For each incoming loop closure, IPC tests compatibility against previously accepted measurements and, if consistent, optimizes only the minimal subgraph and propagates the update; otherwise the loop closure is rejected. The approach achieves competitive precision, recall, and RPE across standard benchmarks and runs online, with an open-source implementation released. This work enables scalable, online robust PGO for SLAM backends and suggests future integration with offline revision to correct past decisions.

Abstract

In SLAM (Simultaneous localization and mapping) problems, Pose Graph Optimization (PGO) is a technique to refine an initial estimate of a set of poses (positions and orientations) from a set of pairwise relative measurements. The optimization procedure can be negatively affected even by a single outlier measurement, with possible catastrophic and meaningless results. Although recent works on robust optimization aim to mitigate the presence of outlier measurements, robust solutions capable of handling large numbers of outliers are yet to come. This paper presents IPC, acronym for Incremental Probabilistic Consensus, a method that approximates the solution to the combinatorial problem of finding the maximally consistent set of measurements in an incremental fashion. It evaluates the consistency of each loop closure measurement through a consensus-based procedure, possibly applied to a subset of the global problem, where all previously integrated inlier measurements have veto power. We evaluated IPC on standard benchmarks against several state-of-the-art methods. Although it is simple and relatively easy to implement, IPC competes with or outperforms the other tested methods in handling outliers while providing online performances. We release with this paper an open-source implementation of the proposed method.

IPC: Incremental Probabilistic Consensus-based Consistent Set Maximization for SLAM Backends

TL;DR

IPC addresses robustness in SLAM back-ends by incrementally building a maximally consistent set of loop closures using a consensus mechanism. For each incoming loop closure, IPC tests compatibility against previously accepted measurements and, if consistent, optimizes only the minimal subgraph and propagates the update; otherwise the loop closure is rejected. The approach achieves competitive precision, recall, and RPE across standard benchmarks and runs online, with an open-source implementation released. This work enables scalable, online robust PGO for SLAM backends and suggests future integration with offline revision to correct past decisions.

Abstract

In SLAM (Simultaneous localization and mapping) problems, Pose Graph Optimization (PGO) is a technique to refine an initial estimate of a set of poses (positions and orientations) from a set of pairwise relative measurements. The optimization procedure can be negatively affected even by a single outlier measurement, with possible catastrophic and meaningless results. Although recent works on robust optimization aim to mitigate the presence of outlier measurements, robust solutions capable of handling large numbers of outliers are yet to come. This paper presents IPC, acronym for Incremental Probabilistic Consensus, a method that approximates the solution to the combinatorial problem of finding the maximally consistent set of measurements in an incremental fashion. It evaluates the consistency of each loop closure measurement through a consensus-based procedure, possibly applied to a subset of the global problem, where all previously integrated inlier measurements have veto power. We evaluated IPC on standard benchmarks against several state-of-the-art methods. Although it is simple and relatively easy to implement, IPC competes with or outperforms the other tested methods in handling outliers while providing online performances. We release with this paper an open-source implementation of the proposed method.
Paper Structure (12 sections, 7 equations, 4 figures, 2 tables, 1 algorithm)

This paper contains 12 sections, 7 equations, 4 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. Assumptions and Heuristics
  5. Problem Formulation
  6. Independent Subgraphs and Solution Propagation
  7. The IPC Algorithm
  8. Experiments
  9. Datasets
  10. Metrics
  11. Evaluation
  12. Conclusions

Figures (4)

  • Figure 1: Trajectory estimated by the different methods on the FR079 dataset corrupted with 70% of outliers.
  • Figure 2: An example of pose graph with nodes represented by light blue circles. Each error term is associated with the corresponding edge. The coordinate system is usually fixed in the first node $\mathbf{x}_0$. Light green and purple ovals represent examples of simple independent subgraphs (i.e., they only include one loop) while the dotted red oval represents more complex independent subgraphs, that include both multiple loops with crossing edges and internal loops.
  • Figure 3: Top left, top right, and bottom images are respectively the precision, the recall, and F1 score performances of the methods for varying outlier percentages. Legend of the graphs: HUBER (), GM(), IPC(), DCS(), MAXMIX(), ADAPT(), PCM(), GNC(). GM is completely covered by DCS in both precision and recall plots, while MAXMIX is nearly entirely covered by IPC in the precision plot.
  • Figure 4: Top left, top right, and bottom images are respectively the ATE, the RPE, and convergence time of the methods for varying outlier percentages. Legend of the graphs: HUBER (), GM(), IPC(), DCS(), MAXMIX(), ADAPT(), PCM(), GNC(). The MAXMIX plot for the RPE metrics is totally covered by the ADAPT plot.