Global Pose Estimation with an Attention-based Recurrent Network

TL;DR

This paper introduces a differentiable neural network architecture for SLAM that learns to perform local pose estimation, pose selection, and graph optimization end-to-end, demonstrated in simulated environments.

Contribution

The novel Neural Graph Optimizer architecture integrates pose estimation, selection, and optimization into a unified trainable system for SLAM.

Findings

Effective in simulated 2D maze environment

Successfully applied to 3D ViZ-Doom environment

End-to-end training improves domain-specific feature learning

Abstract

The ability for an agent to localize itself within an environment is crucial for many real-world applications. For unknown environments, Simultaneous Localization and Mapping (SLAM) enables incremental and concurrent building of and localizing within a map. We present a new, differentiable architecture, Neural Graph Optimizer, progressing towards a complete neural network solution for SLAM by designing a system composed of a local pose estimation model, a novel pose selection module, and a novel graph optimization process. The entire architecture is trained in an end-to-end fashion, enabling the network to automatically learn domain-specific features relevant to the visual odometry and avoid the involved process of feature engineering. We demonstrate the effectiveness of our system on a simulated 2D maze and the 3D ViZ-Doom environment.