This Zenodo record contains the English and Japanese editions of the AutoE2E technical paper, together with Vivliostyle/LaTeX source, publication figures, reproducibility scripts, and audit evidence. AutoE2E is a BEVFormer-V2-based temporal BEV planner that fuses separately encoded HD-map and route rasters with camera BEV features and predicts acceleration and curvature through a recurrent planner. The reported evaluation covers KITScenes Val v3.5 under Camera + HD Map + oracle post-hoc Route inputs and a separate invalid-navigation-input KITScenes Test v1.0 track. Scope limitation: the KITScenes route is reconstructed retrospectively from the logged whole-scene trajectory. Results are open-loop. The Camera-only Test population differs from Val and is not a causal Map/Route ablation. No closed-loop safety, intervention, production-readiness, or state-of-the-art claim is made.
No takes yet. Share an insight, caveat, or question.
Ryota Yamada (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: