AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

KING: Generating Safety-Critical Driving Scenarios for Robust Imitation via Kinematics Gradients

Niklas Hanselmann, Katrin Renz, Kashyap Chitta, Apratim Bhattacharyya, Andreas Geiger

ECCV 2022 · Image generation › Autoregressive and latent generation · approach: Diffusion planning and policies · 115 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
stop gradients through the indirect path (policy and renderer), use only the direct kinematics path (ablation)KING · CARLA scenario generation (overall)Collision rate: 75.83 -> 82.50Improved
stop gradients through the indirect path, use only the direct path (ablation)KING · CARLA scenario generation (overall)Time to 50% collision rate: 16.14 -> 7.78Improved
stop gradients through the indirect path, use only the direct path (ablation)KING · CARLA scenario generation (overall)Seconds per iteration: 3.27 -> 1.90Improved
replace black-box optimization (CMA-ES) with kinematics-gradient optimizationAIM-BEV attack · CARLA scenario generation (overall)Collision rate: 68.33 -> 82.50Improved
increase number of adversarial agents from 1 to 4KING · CARLA scenario generationCollision rate: 86.25 -> 78.75Hurt
increase number of adversarial agents from 1 to 2 then 4CMA-ES · CARLA scenario generationCollision rate: 67.50 -> 75.00 -> 62.50Mixed
fine-tune on regular data D_reg onlyAIM-BEV · Held-out KING scenariosCollision rate: 100.00 -> 57.14Improved
fine-tune on regular data D_reg onlyAIM-BEV · Town10 intersectionsCollision rate: 17.48 -> 19.51Hurt
add safety-critical data D_crit to regular data D_reg in fine-tuning (ablation)AIM-BEV · Held-out KING scenariosCollision rate: 57.14 -> 28.57Improved
add safety-critical data D_crit to regular data D_reg in fine-tuning (ablation)AIM-BEV · Town10 intersectionsCollision rate: 19.51 -> 8.13Improved
fine-tune on D_crit + D_reg instead of no fine-tuningAIM-BEV · Town10 intersectionsDriving Score: 86.74 -> 90.20Improved
add regular data D_reg to D_crit in fine-tuning (ablation)AIM-BEV · Town10 intersectionsDriving Score: 88.37 -> 90.20Improved

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