AI Paper Atlas

Research map / Probabilistic and scientific modeling / Treatment effects

CausalVAD: De-confounding End-to-End Autonomous Driving via Causal Intervention

Jiacheng Tang, Zhiyuan Zhou, Zhuolin He, Jia Zhang, Kai Zhang, Jian Pu

CVPR 2026 · Probabilistic and scientific modeling › Treatment effects · approach: Proximal and deep causal effect estimation · 6 citations (OpenAlex)

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
add Perception de-confounding module (PDM) (ablation)VAD-tiny · nuScenes valavg collision rate (%): 0.44 -> 0.26Improved
add Interaction de-confounding module (IDM) (ablation)VAD-tiny · nuScenes valavg L2 (m): 0.74 -> 0.57Improved
add both PDM and IDM (ablation)VAD-tiny · nuScenes valavg L2 (m): 0.74 -> 0.54Improved
use dictionary sizes (10,3,6) instead of smaller (5,2,3) (ablation)CausalVAD · nuScenes valavg L2 (m): 0.58 -> 0.54Improved
use larger dictionary (20,5,10) instead of (10,3,6) (ablation)CausalVAD · nuScenes valavg L2 (m): 0.54 -> 0.56Hurt
swap K-means++ for K-means or K-medoids clustering (ablation)CausalVAD · nuScenes valavg L2 (m): 0.54 vs 0.53 / 0.54No clear effect
apply causal de-confoundingVAD-tiny · nuScenes turning scenariosavg L2 (m): 1.07 -> 0.69Improved
apply causal de-confoundingVAD-tiny · nuScenes ego-velocity zeroedavg L2 (m): 6.94 -> 4.80Improved
apply causal de-confoundingVAD-tiny · nuScenes ego-velocity 100m/savg collision rate (%): 11.62 -> 5.19Improved
apply causal de-confoundingVAD-tiny · nuScenes ego-velocity x0.5avg L2 (m): 3.80 -> 2.84Improved
apply causal de-confoundingVAD-tiny · nuScenes valavg collision rate (%): 0.44 -> 0.11Improved
apply causal de-confoundingVAD-tiny · NAVSIMPDMS: 80.5 -> 87.2Improved

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