Research map / 3D vision / Depth and stereo
BAD SLAM: Bundle Adjusted Direct RGB-D SLAM
Thomas Schops, Torsten Sattler, Marc Pollefeys
Check your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add camera intrinsics and depth deformation optimization (ablation) | BAD SLAM · TUM RGB-D | ATE RMSE: 3.6 -> 1.7; 1.2 -> 1.1; 2.5 -> 1.7 | Improved |
| add geometric residuals to photometric residuals (ablation) | BAD SLAM · ETH3D benchmark training datasets | ATE RMSE | Improved |
| add photometric residuals to depth-only geometric residuals (ablation) | BAD SLAM · ETH3D benchmark training datasets | ATE RMSE | Improved |
| run the BA back-end in addition to the front-end (ablation) | BAD SLAM · ETH3D benchmark training datasets | ATE RMSE | Improved |
| run 25 BA iterations after dataset playback and do not skip iterations (ablation) | BAD SLAM · ETH3D benchmark training datasets | ATE RMSE | Improved |
| replace alternating optimization with a preconditioned conjugate gradient solver (ablation) | BAD SLAM · ETH3D benchmark training datasets | ATE RMSE | Hurt |
| select keyframes more frequently (ablation) | BAD SLAM · ETH3D benchmark training datasets | ATE RMSE | Improved |
| increase surfel creation cell size beyond 8 × 8 (ablation) | BAD SLAM · ETH3D benchmark training datasets | ATE RMSE | Hurt |
| use 2 × 2 to 8 × 8 surfel creation cells rather than varying within this range (ablation) | BAD SLAM · ETH3D benchmark training datasets | ATE RMSE | No clear effect |
| record synchronized global-shutter RGB-D data instead of data affected by asynchronous frames or rolling shutter | direct RGB-D SLAM methods · Synthetic RGB-D datasets | ATE RMSE | Hurt |
| use BAD SLAM rather than ORB-SLAM2 | RGB-D SLAM · ETH3D benchmark | ATE RMSE | Improved |
| use BAD SLAM rather than ORB-SLAM2 | RGB-D SLAM · TUM RGB-D | ATE RMSE | Hurt |
Later papers that compared with this method
2 later accepted papers report BAD SLAM as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Gradient-SDF: A Semi-Implicit Surface Representation for 3D Reconstruction (CVPR 2022 · better on 3 of 5 shared results)
- iMAP: Implicit Mapping and Positioning in Real-Time (ICCV 2021 · better on 0 of 3 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- VGGT-SLAM: Dense RGB SLAM Optimized on the SL(4) Manifold (NeurIPS 2025 · similarity 0.87)
- Loopy-SLAM: Dense Neural SLAM with Loop Closures (CVPR 2024 · similarity 0.87)
- GS-SLAM: Dense Visual SLAM with 3D Gaussian Splatting (CVPR 2024 · similarity 0.87)
- WildGS-SLAM: Monocular Gaussian Splatting SLAM in Dynamic Environments (CVPR 2025 · similarity 0.87)
- Photo-SLAM: Real-time Simultaneous Localization and Photorealistic Mapping for Monocular Stereo and RGB-D Cameras (CVPR 2024 · similarity 0.87)
- SLAM3R: Real-Time Dense Scene Reconstruction from Monocular RGB Videos (CVPR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
