DROID-SLAM: Deep Visual SLAM for Monocular, Stereo, and RGB-D Cameras

TL;DR

DROID-SLAM is a deep learning-based SLAM system that iteratively updates camera pose and depth, achieving high accuracy and robustness across monocular, stereo, and RGB-D videos.

Contribution

It introduces a novel recurrent iterative update mechanism with a Dense Bundle Adjustment layer for improved SLAM performance.

Findings

Significantly outperforms prior SLAM methods in accuracy.

Exhibits fewer catastrophic failures, demonstrating robustness.

Can leverage stereo and RGB-D data even when trained on monocular videos.

Abstract

We introduce DROID-SLAM, a new deep learning based SLAM system. DROID-SLAM consists of recurrent iterative updates of camera pose and pixelwise depth through a Dense Bundle Adjustment layer. DROID-SLAM is accurate, achieving large improvements over prior work, and robust, suffering from substantially fewer catastrophic failures. Despite training on monocular video, it can leverage stereo or RGB-D video to achieve improved performance at test time. The URL to our open source code is https://github.com/princeton-vl/DROID-SLAM.