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

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.

References in corpus (2)

• DeepVO: Towards End-to-End Visual Odometry with Deep Recurrent Convolutional Neural Networks
• TartanVO: A Generalizable Learning-based VO

Cited by in corpus (16)

• Vox-Fusion: Dense Tracking and Mapping with Voxel-based Neural Implicit Representation
• Autonomous Drone Racing: A Survey
• Visual SLAM: What are the Current Trends and What to Expect?
• Hilti-Oxford Dataset: A Millimetre-Accurate Benchmark for Simultaneous Localization and Mapping
• PyPose: A Library for Robot Learning with Physics-based Optimization
• VIO-DualProNet: Visual-Inertial Odometry with Learning Based Process Noise Covariance
• Benchmarking Neural Radiance Fields for Autonomous Robots: An Overview
• BAMF-SLAM: Bundle Adjusted Multi-Fisheye Visual-Inertial SLAM Using Recurrent Field Transforms
• RDFC-GAN: RGB-Depth Fusion CycleGAN for Indoor Depth Completion
• RGB-Only Gaussian Splatting SLAM for Unbounded Outdoor Scenes
• SplatMAP: Online Dense Monocular SLAM with 3D Gaussian Splatting
• Leveraging Cycle-Consistent Anchor Points for Self-Supervised RGB-D Registration
• CodedVO: Coded Visual Odometry
• UDGS-SLAM : UniDepth Assisted Gaussian Splatting for Monocular SLAM
• InCrowd-VI: A Realistic Visual-Inertial Dataset for Evaluating SLAM in Indoor Pedestrian-Rich Spaces for Human Navigation
• ConfidentSplat: Confidence-Weighted Depth Fusion for Accurate 3D Gaussian Splatting SLAM