Research map / Label-efficient and robust learning / Recognition and long-tail
PriorEye: Geospatial Visual Priors for End-to-End Autonomous Driving
Kyuhwan Yeon, Benjamin Ramtoula, Daniele De Martini
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 PriorEye memory augmentation | LTF · NAVSIM-v2 navhard-two-stage | EPDMS: 24.7 -> 32.4 | Improved |
| add PriorEye memory augmentation | GTRS-DP · NAVSIM-v2 navhard-two-stage | EPDMS: 26.3 -> 30.1 | Improved |
| add PriorEye memory augmentation | GTRS-Dense · NAVSIM-v2 navhard-two-stage | EPDMS: 44.9 -> 48.6 | Improved |
| add PriorEye memory augmentation | DrivoR · NAVSIM-v2 navhard-two-stage | EPDMS: 48.9 -> 49.6 | Improved |
| add PriorEye memory augmentation | GTRS-DP · NAVSIM-v2 navtest | EPDMS: 82.2 -> 82.5 | No clear effect |
| add PriorEye memory augmentation under heavy mud corruption | GTRS-Dense · navhard-two-stage, sensor corruption: Mud (Heavy) | EPDMS: 21.0 -> 32.8 | Improved |
| add persistent memory P to contextual memory C under visual prior corruption (ablation) | PriorEye · navhard-two-stage, visual prior corruption | EPDMS: 38.5 -> 45.3 | Improved |
| use intention-guided retrieval instead of proximity-based retrieval (ablation) | PriorEye (LTF) · navtrain held-out validation | EPDMS: 82.1 -> 84.4 | Improved |
| add contextual memory C to persistent memory P only (ablation) | PriorEye (LTF) · navtrain held-out validation | EPDMS: 82.2 -> 84.4 | Improved |
| add persistent memory P to contextual memory C (ablation) | PriorEye (LTF) · navtrain held-out validation | EPDMS: 83.9 -> 84.4 | Improved |
| add visual encoding to spatial-only contextual memory (ablation) | PriorEye (LTF) · navtrain held-out validation | EPDMS: 82.7 -> 84.4 | Improved |
| add distance and cosine cues to the memory gate (ablation) | PriorEye (LTF) · navtrain held-out validation | EPDMS: 83.0 -> 84.4 | Improved |
Closest papers by meaning
- FutureSightDrive: Thinking Visually with Spatio-Temporal CoT for Autonomous Driving (NeurIPS 2025 · similarity 0.88)
- SpaceDrive: Infusing Spatial Awareness into VLM-based Autonomous Driving (CVPR 2026 · similarity 0.87)
- Future-Aware End-to-End Driving: Bidirectional Modeling of Trajectory Planning and Scene Evolution (NeurIPS 2025 · similarity 0.87)
- LaST-VLA: Thinking in Latent Spatio-Temporal Space for Vision-Language-Action in Autonomous Driving (NeurIPS 2026 · similarity 0.86)
- DriveSpatial: A Benchmark for Spatiotemporal Intelligence in VLMs for Autonomous Driving (NeurIPS 2026 · similarity 0.86)
- VR-Drive: Viewpoint-Robust End-to-End Driving with Feed-Forward 3D Gaussian Splatting (NeurIPS 2025 · similarity 0.86)
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.
