Research map / 3D vision / Depth and stereo
Benchmarking Egocentric Visual-Inertial SLAM at City Scale
Anusha Krishnan, Shaohui Liu, Paul-Edouard Sarlin, Oscar Gentilhomme, David Caruso, Maurizio Monge, Richard Newcombe, Jakob Engel, Marc Pollefeys
Read the paperCheck 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 |
|---|---|---|---|
| use multiple cameras plus inertial instead of mono or mono+inertial | OpenVINS · LaMAria main (short) | score: 18.1 -> 22.2 | Improved |
| add Maplab offline visual-inertial optimization on top of OpenVINS (ablation) | OpenVINS mono+inertial · LaMAria main (short) | score: 18.1 -> 22.9 | Improved |
| add Maplab on top of multi-camera OpenVINS (ablation) | OpenVINS multi-cam+inertial · LaMAria main (short) | score: 22.2 -> 26.0 | Improved |
| move from controlled platform motion (level I) to egocentric motion (level IV) (ablation) | most VIO/SLAM systems · LaMAria controlled set | ATE RMSE / failures: DPVO 0.22 -> 42.07 (seq 1 vs seq 11) | Hurt |
| add online calibration, full VI bundle adjustment and non-causal processing (Aria's SLAM) versus best academic system | multi-cam+inertial SLAM · LaMAria main (long) | score: 12.6 -> 70.8 | Improved |
| moving platform (tram, funicular) condition | Aria's SLAM · LaMAria main | score: 70.8 -> 53.6 (long vs moving platform) | Hurt |
| use offline SfM (GLOMAP) instead of monocular ORB-SLAM3 on short snippets | outdoor walking snippets · LaMAria short snippets (outdoor walking) | R@50cm: 31.3 -> 64.1 | Improved |
| add inertial to visual methods under low light and exposure change | ORB-SLAM3 / OpenVINS vs COLMAP · LaMAria short snippets (low light) | R@50cm: 7.5 -> 22.2 (COLMAP vs OpenVINS mono+inertial) | Improved |
| use more than one camera for the long sequences under moving platform | OpenVINS · LaMAria main (moving platform) | score: 2.4 -> 11.5 | Improved |
| evaluate in 3D instead of 2D | scoring of all systems · LaMAria main | score: Aria's SLAM short 90.7 -> 87.7 | Hurt |
Closest papers by meaning
- ADVIO: An Authentic Dataset for Visual-Inertial Odometry (ECCV 2018 · similarity 0.88)
- Long-Term Visual Localization Revisited (TPAMI 2020 · similarity 0.88)
- Benchmarking 6DOF Outdoor Visual Localization in Changing Conditions (CVPR 2018 · similarity 0.88)
- 4Seasons: Benchmarking Visual SLAM and Long-Term Localization for Autonomous Driving in Challenging Conditions (IJCV 2024 · similarity 0.87)
- UniSim-SLAM: Feed-Forward SLAM with Unified Sim(3) Optimization (ECCV 2026 · similarity 0.87)
- RoboSense: Large-scale Dataset and Benchmark for Egocentric Robot Perception and Navigation in Crowded and Unstructured Environments (CVPR 2025 · similarity 0.85)
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.
