DROID-SLAM in the Wild
Not on the map. Why?
Abstract (arXiv)
We present a robust, real-time RGB SLAM system that handles dynamic environments by leveraging differentiable Uncertainty-aware Bundle Adjustment. Traditional SLAM methods typically assume static scenes, leading to tracking failures in the presence of motion. Recent dynamic SLAM approaches attempt to address this challenge using predefined dynamic priors or uncertainty-aware mapping, but they remain limited when confronted with unknown dynamic objects or highly cluttered scenes where geometric mapping becomes unreliable. In contrast, our method estimates per-pixel uncertainty by exploiting multi-view visual feature inconsistency, enabling robust tracking and reconstruction even in real-world environments. The proposed system achieves state-of-the-art camera poses and scene geometry in cluttered dynamic scenarios while running in real time at around 10 FPS. Code and datasets are available at https://github.com/MoyangLi00/DROID-W.git.
Key facts. Cited by 9 papers in the map (data through 7 Oct 2026). Builds on 55 of its 72 references that are in the map. First posted on arXiv 19 Mar 2026.
Builds on
55 of 72 references.
References · 55
- 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,764
- 2017Mask R-CNNcited by 7,259
- 20233D Gaussian Splatting for Real-Time Radiance Field Renderingcited by 6,141
- 2015SSD: Single Shot MultiBox Detectorcited by 3,125
- 2023DUSt3R: Geometric 3D Vision Made Easycited by 1,526
- 2015SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentationcited by 1,483
- 2025VGGT: Visual Geometry Grounded Transformercited by 1,475
- 2015ORB-SLAM: a Versatile and Accurate Monocular SLAM Systemcited by 983
- 2016ORB-SLAM2: an Open-Source SLAM System for Monocular, Stereo and RGB-D Camerascited by 801
- 2021DROID-SLAM: Deep Visual SLAM for Monocular, Stereo, and RGB-D Camerascited by 673
- 2021NICE-SLAM: Neural Implicit Scalable Encoding for SLAMcited by 586
- 2016Direct Sparse Odometrycited by 479
38 more
- 2021iMAP: Implicit Mapping and Positioning in Real-Timecited by 476
- 2024MonST3R: A Simple Approach for Estimating Geometry in the Presence of Motioncited by 431
- 2023Gaussian Splatting SLAMcited by 410
- 2024Metric3Dv2: A Versatile Monocular Geometric Foundation Model for Zero-shot Metric Depth and Surface Normal Estimationcited by 389
- 2023SplaTAM: Splat, Track & Map 3D Gaussians for Dense RGB-D SLAMcited by 386
- 2023GS-SLAM: Dense Visual SLAM with 3D Gaussian Splattingcited by 278
- 2024MegaSaM: Accurate, Fast, and Robust Structure and Motion from Casual Dynamic Videoscited by 224
- 2022Monocular Dynamic View Synthesis: A Reality Checkcited by 199
- 2025EgoDex: Learning Dexterous Manipulation from Large-Scale Egocentric Videocited by 191
- 2022Deep Patch Visual Odometrycited by 186
- 2023Co-SLAM: Joint Coordinate and Sparse Parametric Encodings for Neural Real-Time SLAMcited by 183
- 2019ReFusion: 3D Reconstruction in Dynamic Environments for RGB-D Cameras Exploiting Residualscited by 180
- 2023NICER-SLAM: Neural Implicit Scene Encoding for RGB SLAMcited by 162
- 2023Point-SLAM: Dense Neural Point Cloud-based SLAMcited by 160
- 2022ESLAM: Efficient Dense SLAM System Based on Hybrid Representation of Signed Distance Fieldscited by 158
- 2024WildGaussians: 3D Gaussian Splatting in the Wildcited by 150
- 2022Vox-Fusion: Dense Tracking and Mapping with Voxel-based Neural Implicit Representationcited by 149
- 2019Self-Driving Cars: A Surveycited by 147
- 2023GO-SLAM: Global Optimization for Consistent 3D Instant Reconstructioncited by 132
- 2018DynaSLAM: Tracking, Mapping and Inpainting in Dynamic Scenescited by 130
- 2025TTT3R: 3D Reconstruction as Test-Time Trainingcited by 122
- 2018MaskFusion: Real-Time Recognition, Tracking and Reconstruction of Multiple Moving Objectscited by 95
- 2018DS-SLAM: A Semantic Visual SLAM towards Dynamic Environmentscited by 90
- 2025Easi3R: Estimating Disentangled Motion from DUSt3R Without Trainingcited by 88
- 2024NeRF On-the-go: Exploiting Uncertainty for Distractor-free NeRFs in the Wildcited by 85
- 2024Improving 2D Feature Representations by 3D-Aware Fine-Tuningcited by 70
- 2024RGBD GS-ICP SLAMcited by 65
- 2024Splat-SLAM: Globally Optimized RGB-only SLAM with 3D Gaussianscited by 64
- 2024CG-SLAM: Efficient Dense RGB-D SLAM in a Consistent Uncertainty-aware 3D Gaussian Fieldcited by 61
- 2023How To Not Train Your Dragon: Training-free Embodied Object Goal Navigation with Semantic Frontierscited by 53
- 2025WildGS-SLAM: Monocular Gaussian Splatting SLAM in Dynamic Environmentscited by 48
- 2017Co-Fusion: Real-time Segmentation, Tracking and Fusion of Multiple Objectscited by 47
- 2024FAST-LIVO2: Fast, Direct LiDAR-Inertial-Visual Odometrycited by 40
- 2020FlowFusion: Dynamic Dense RGB-D SLAM Based on Optical Flowcited by 36
- 2024DG-SLAM: Robust Dynamic Gaussian Splatting SLAM with Hybrid Pose Optimizationcited by 29
- 2024RoDyn-SLAM: Robust Dynamic Dense RGB-D SLAM with Neural Radiance Fieldscited by 22
- 2023DynaMoN: Motion-Aware Fast and Robust Camera Localization for Dynamic Neural Radiance Fieldscited by 16
- 2023Scientific Exploration of Challenging Planetary Analog Environments with a Team of Legged Robotscited by 14
50 most cited of 55 All 55 (JSON)
Built on by
Citing papers per year: 2026: 9.
9 papers.
- 2026Open-AoE: An Open Egocentric Manipulation Dataset and Toolchain for Embodied Learningcited by 5
- 2026VidMap: Exploiting Temporal Structure for Video-Based Structure-from-Motioncited by 2
- 2026OpenLongTail: Generative Scaling of Long-Tail Driving Datacited by 1
- 2026Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depthcited by 0
- 2026Temporally Grounded Compositional Camera Motion Understanding via Geometric Knowledge Distillationcited by 0
- 2026Revisiting Local Context for Long-Horizon Streaming 3D Reconstructioncited by 0
- 2026SCOUT-SLAM: Structurally-Coupled Dual Uncertainty-Aware 3DGS SLAM in the Wildcited by 0
- 2026MEgoVista: Multi-view Ego-aware Motion Estimation for Metric 4D Hands and Head in the Wildcited by 0
- 2026CLoSeR: Closing the Loop for Long-Context Streaming Reconstructioncited by 0
- 2026CLoSeR: Closing the Loop for Long-Context Streaming Reconstruction1 Oct 2026
- 2026MEgoVista: Multi-view Ego-aware Motion Estimation for Metric 4D Hands and Head in the Wild15 Sep 2026
- 2026SCOUT-SLAM: Structurally-Coupled Dual Uncertainty-Aware 3DGS SLAM in the Wild13 Sep 2026
- 2026Revisiting Local Context for Long-Horizon Streaming 3D Reconstruction27 Aug 2026
- 2026Temporally Grounded Compositional Camera Motion Understanding via Geometric Knowledge Distillation11 Aug 2026
- 2026VidMap: Exploiting Temporal Structure for Video-Based Structure-from-Motion29 Jul 2026
- 2026Open-AoE: An Open Egocentric Manipulation Dataset and Toolchain for Embodied Learning15 Jul 2026
- 2026OpenLongTail: Generative Scaling of Long-Tail Driving Data10 Jul 2026
- 2026Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth19 May 2026