Research map / 3D vision / 3D generation
UniMapping: Unified SLAM Framework for Map-Centric Embodied Perception
Xiaze Zhang, Ziheng Ding, Yuejie Zhang, lifeng chen, Rui Feng
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 |
|---|---|---|---|
| add Spatial Fusion to UniMapping detection (ablation) | UniMapping · ScanNet detection | mAP25: 57.7 -> 60.8 | Improved |
| add Spatial Fusion to UniMapping segmentation (ablation) | UniMapping · SemanticKITTI | mIoU: 53.7 -> 60.3 | Improved |
| use Temporal Fusion (10-frame window) instead of Spatial Fusion (ablation) | UniMapping · SemanticKITTI | mIoU: 53.7 -> 53.8 | No clear effect |
| use Spatial Fusion instead of Temporal Fusion (ablation) | UniMapping · ScanNet detection | mAP25: 59.2 -> 60.8 | Improved |
| replace cross attention with SADT in descriptor extraction (ablation) | UniMapping · ScanNet SLAM seq 000 (25% train) | ATE (cm): 21.6 -> 13.4 | Improved |
| replace vanilla deformable attention with SADT (ablation) | UniMapping · ScanNet SLAM seq 059 | ATE (cm): 27.1 -> 14.9 | Improved |
| use SADT instead of simple interpolation (ablation) | UniMapping · ScanNet SLAM seq 000 | ATE (cm): 14.0 -> 13.4 | Improved |
| use cross attention for spatial fusion instead of none (ablation) | UniMapping · ScanNet detection (25% train) | mAP25: 44.09 -> 41.29 | Hurt |
| use multi-scale spatial aggregation instead of deformable attention for fusion (ablation) | UniMapping · ScanNet detection (25% train) | mAP25: 46.27 -> 48.84 | Improved |
| add pose noise of 15 cm to Spatial Fusion (ablation) | UniMapping · ScanNet detection | mAP25: decreases by 1.2% absolute | No clear effect |
| distance-based map pruning (5 m) (ablation) | UniMapping · map memory | map size, ATE: 61 MB -> 31 MB; ATE up approx 8% | Mixed |
| use UniMapping instead of best specialized SLAM on ScanNet seq 000 | UniMapping · ScanNet SLAM | ATE (cm): 4.2 -> 13.7 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 3 on most shared results.
- Loopy-SLAM — Loopy-SLAM: Dense Neural SLAM with Loop Closures (CVPR 2024) · better on 0 of 1 shared result
- ESLAM — ESLAM: Efficient Dense SLAM System Based on Hybrid Representation of Signed Distance Fields (CVPR 2023) · better on 1 of 1 shared result
- GO-SLAM — GO-SLAM: Global Optimization for Consistent 3D Instant Reconstruction (ICCV 2023) · better on 0 of 1 shared result
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
- Benchmarking Implicit Neural Representation and Geometric Rendering in Real-Time RGB-D SLAM (CVPR 2024 · similarity 0.86)
- NICE-SLAM: Neural Implicit Scalable Encoding for SLAM (CVPR 2022 · similarity 0.86)
- Cross-Descriptor Visual Localization and Mapping (ICCV 2021 · similarity 0.86)
- MAGiC-SLAM: Multi-Agent Gaussian Globally Consistent SLAM (CVPR 2025 · similarity 0.86)
- ERGeoBench: A Comprehensive Benchmark for Embodied Reasoning and Geo-localization in Multimodal Large Language Models (ICML 2026 · similarity 0.86)
- Photo-SLAM: Real-time Simultaneous Localization and Photorealistic Mapping for Monocular Stereo and RGB-D Cameras (CVPR 2024 · 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.
