AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

Seeing Beyond the Next Step: World-Model-Guided Human-Like Navigation in Multi-Agent Scenes

SEONGEUN HONG, JuYeong Hwang, Jinhyun Kim, Hanyoung Jang, Hyeongyeop Kang

NeurIPS 2026 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Vision-language navigation agents · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes a world-model-guided navigation approach that evaluates candidate next actions by predicting their consequences in populated multi-agent scenes, to produce more human-like motion.

Problem

Choosing a next motion in populated scenes requires anticipating what each candidate action will cause, since failures such as drifting onto the road, cutting into pedestrian streams, or bottlenecking passages are invisible from the current state alone.

Approach

Uses a world model to roll out the consequences of candidate actions so the planner can reason about downstream effects on other agents and the scene before acting, rather than only avoiding immediate obstacles.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
remove both world models from preference refinement (ablation)Ours · 5 held-out scenes avgGSR / SRVR: GSR 77.4 -> 37.3; SRVR 4.0 -> 19.5Hurt
add state transition world model (ablation)Ours · ETH_IntersectionGSR: 51.9 -> 83.3Improved
add semantic transition world model (ablation)Ours · ETH_HallwaySRVR: 13.7 -> 0.9Improved
use mixed top-5 + random-5 candidates instead of top-10 only (ablation)Ours · Zara01_GroupGSR: 14.8 -> 77.8Improved
use mixed top-5 + random-5 candidates instead of random-10 only (ablation)Ours · ETH_IntersectionGSR: 44.4 -> 83.3Improved
use pretrained VLM backbone instead of task-trained Transformer/ViT (ablation)Ours · Zara01GSR: 48.1 -> 72.2Improved
distill world-model preferences into policy instead of running world models onlineOurs vs OnlineWM · 5 held-out scenes avgGSR: 80.6 (OnlineWM) vs 77.4No clear effect
distilled policy vs online planner inference latencyOurs vs OnlineWM · per decisionlatency: 1.2 s -> 65 msImproved
use world-model-guided policy instead of trajectory predictors with MPCOurs · 5 held-out scenes avgGSR: 56.1 (Constant Velocity) -> 77.4Improved
use world-model-guided policy instead of GBM baseline on group distanceOurs · Zara01_GroupGD: 1.9 (GBM) vs 2.1Hurt
replace baselines with world-model-guided policyOurs · Zara01_GroupSRVR: 17.4 -> 1.2Improved

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