PriorEye: Geospatial Visual Priors for End-to-End Autonomous Driving (ECCV 2026)
rsasaki0109 · x · 2026-08-23
Addressing the limitation of end-to-end autonomous driving methods that rely solely on instantaneous sensor observations, this paper (ECCV 2026) introduces PriorEye. It leverages geospatial visual priors—street-level visual context anchored to the planned route—to provide visual-spatial foresight similar to human driver experience.
Technical Contributions:
- Dual-Memory Architecture: Combines a contextual memory for retrieved priors with a persistent fallback memory.
- Adaptive Memory Gate: Dynamically regulates the influence of memories based on compatibility with the current driving state.
- Seamless Integration: The memory augmentation module is designed to be easily integrated into existing end-to-end approaches.
Performance:
Evaluated on the NAVSIM-v2 benchmark, the approach demonstrates significant improvements in driving performance by incorporating anticipatory prior knowledge.
More from Embodied
- Would You Swallow a Robot to Avoid a Colonoscopy? — TinfoilTricorn · 2026-08-23
- Two-month-old startup's humanoid robot drives a go-kart at full speed on a real track — 量子位 · 2026-08-23
- RoboCup Goal: Beat Human World Cup Champions by 2050 — mathemagic1an · 2026-08-23
- Japanese student team ARES tops ERC prelims, aims for world championship — 4310sy · 2026-08-23
- User experiences Coco robot delivery for the first time — yungcontent · 2026-08-23
- Osaka Metro Univ AI replicates captain steering, follows unlearned rules — 4310sy · 2026-08-23