AI Paper Atlas

Research map / Large language models / Agents, code and math

KnowVal: A Knowledge-Augmented and Value-Guided Autonomous Driving System

Zhongyu Xia, Wenhao Chen, Yongtao Wang, Ming-Hsuan Yang

CVPR 2026 · Large language models › Agents, code and math · approach: LLM-driven embodied and driving agents · 3 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 knowledge retrieval and value model (ablation)GenAD · nuScenescollision rate: 0.43 -> 0.38Improved
add open-world perception (ablation)GenAD + K&V · nuScenescollision rate: 0.38 -> 0.34Improved
add retrieval-guided supplementary perception (ablation)GenAD + K&V + OW · nuScenescollision rate: 0.34 -> 0.33Improved
add knowledge retrieval alone (no value model) (ablation)SimLingo · Bench2DriveDriving Score: 85.07 -> 84.91No clear effect
add learned value model alone (ablation)SimLingo · Bench2DriveDriving Score: 85.07 -> 85.92Improved
add knowledge to value model (ablation)SimLingo + values · Bench2DriveDriving Score: 85.92 -> 87.61Improved
use learned value model instead of heuristic rules (ablation)SimLingo + values · Bench2DriveDriving Score: 85.94 -> 85.92No clear effect
add open-world perception (ablation)KnowVal-SimLingo · Bench2DriveDriving Score: 87.61 -> 88.16Improved
add retrieval-guided supplementary perception (ablation)KnowVal-SimLingo · Bench2DriveDriving Score: 88.16 -> 88.42Improved
insert random items among retrieved knowledge (ablation)KnowVal-SimLingo · Bench2DriveDriving Score: 87.61 -> 87.07Hurt
replace retrieval with random knowledge (ablation)KnowVal-SimLingo · Bench2DriveDriving Score: 87.61 -> 86.18Hurt
use scene features and subscores instead of BEV image with total score (ablation)Value model · value predictionMSE: 0.22 -> 0.11Improved

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