AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

HAD: Combining Hierarchical Diffusion with Metric-Decoupled RL for End-to-End Driving

Wenhao Yao, Xinglong Sun, Zhenxin Li, Shiyi Lan, Zi Wang, Jose M Alvarez, Zuxuan Wu

ECCV 2026 · Reinforcement learning › Visual and representation RL · approach: RL for combinatorial routing

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 changedWhereMeasuredEffect
expand top-K anchors in polar coordinates instead of adding random noise (ablation)HAD · NAVSIM v2EPDMS: 84.9 -> 88.6Improved
expand in polar rather than Cartesian (XY) coordinates (ablation)HAD · NAVSIM v2EPDMS: 85.9 -> 88.6Improved
restrict first-stage candidates to K=2 instead of all 20 (ablation)HAD · NAVSIM v2EPDMS: 79.8 -> 88.6Improved
increase inference expansion density from 5x5 to 7x7 (ablation)HAD · NAVSIM v2EPDMS: +0.1 EPDMSNo clear effect
decouple RL reward across metrics (MDPO) instead of coupled PDMS reward (ablation)HAD · NAVSIM v2EPDMS: 87.8 -> 88.6Improved
use distance-only reward for RL (ablation)HAD · NAVSIM v2EPDMS: 87.8 -> 62.6 (vs PDMS reward)Hurt
add distance to PDMS in a coupled reward (ablation)HAD · NAVSIM v2EPDMS: 87.8 -> 86.9Hurt
add safety metric scoring to refinement stage (ablation)HAD · NAVSIM v2EPDMS: 86.7 -> 87.3Improved
add RL training in the local refinement stage (ablation)HAD · NAVSIM v2EPDMS: 87.3 -> 88.6Improved
add safety and RL supervision to the first (intention) stage (ablation)HAD · NAVSIM v2EPDMS: 88.6 -> 87.2Hurt
replace online simulator rewards with offline reward retrieval (ablation)HAD · training efficiencytraining time (h): 64.4 -> 13.6Improved
drop LiDAR input (camera-only HAD-L) (ablation)HAD · NAVSIM v1PDMS: 90.2 -> 89.9No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy