Self-Supervised Deep Visual Odometry with Online Adaptation

TL;DR

This paper introduces an online meta-learning approach for self-supervised visual odometry that enables continuous adaptation to new environments, significantly improving robustness and accuracy in diverse scenes.

Contribution

It proposes a novel online meta-learning algorithm with convLSTM and feature alignment for self-supervised VO, allowing real-time adaptation to changing environments.

Findings

Outperforms state-of-the-art self-supervised VO methods on unseen scenes.

Demonstrates effective adaptation from virtual to real-world environments.

Achieves consistent improvements in outdoor and indoor scene estimation.

Abstract

Self-supervised VO methods have shown great success in jointly estimating camera pose and depth from videos. However, like most data-driven methods, existing VO networks suffer from a notable decrease in performance when confronted with scenes different from the training data, which makes them unsuitable for practical applications. In this paper, we propose an online meta-learning algorithm to enable VO networks to continuously adapt to new environments in a self-supervised manner. The proposed method utilizes convolutional long short-term memory (convLSTM) to aggregate rich spatial-temporal information in the past. The network is able to memorize and learn from its past experience for better estimation and fast adaptation to the current frame. When running VO in the open world, in order to deal with the changing environment, we propose an online feature alignment method by aligning feature distributions at different time. Our VO network is able to seamlessly adapt to different environments. Extensive experiments on unseen outdoor scenes, virtual to real world and outdoor to indoor environments demonstrate that our method consistently outperforms state-of-the-art self-supervised VO baselines considerably.