Driver-WM: A Driver-Centric Traffic-Conditioned Latent World Model for In-Cabin Dynamics Rollout

arXiv 2026 60.7 method, application

TLDR

Driver-WM is a latent world model that forecasts in-cabin driver dynamics conditioned on external traffic context using a dual-stream architecture with gated causal injection.

Reasoning

The paper introduces a novel driver-centric world model that unifies physical, behavioral, and emotional forecasting, with a strong causal conditioning mechanism. Its strengths include a clear problem formulation and evaluation on a multi-task benchmark, but the abstract lacks details on real-world data sources and comparisons to baselines.

Read-first score

Read-first score 60.7, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 48.

Recency 6%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Methodology quality 18%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=analysis,benchmark,evaluation,metric

Topical relevance 29%

68.6

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Citation impact 18%

68.4

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.68396323

Reproducibility 18%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=code

Citation velocity 12%

Citation velocity estimates citations per publication-year to reduce old-paper bias. velocity=0.00

Field roles

FrontierBridgeMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 386.

Keyword Scores

world model

generative world model

world dynamics prediction

world simulator

interactive world model

model-based reinforcement learning world model

video world model

Deep Analysis

Innovations

Driver-centric latent world model for in-cabin dynamics rollout
Causal conditioning of in-cabin dynamics on out-cabin traffic context
Unified physical kinematics forecasting with auxiliary behavioral and emotional semantic recognition
Dual-stream architecture for separate encoding of external traffic and internal driver states
Gated causal injection mechanism with learned vector gate for directional coupling and temporal causality
Controlled test-time interventions for systematic mechanism analysis

Methodology

Driver-WM operates in a compact latent space constructed from frozen vision-language features. It adopts a dual-stream architecture to separately encode external traffic and internal driver states, directionally coupled via a gated causal injection mechanism that uses a learned vector gate to modulate external contextual perturbations while strictly enforcing temporal causality. The model is evaluated on a multi-task assistive driving benchmark.

Key Results

Driver-WM yields robust long-horizon geometric forecasting for reactive high-motion maneuvers and improves semantic alignment for both driver and traffic states.

Limitations

Evaluation is limited to a multi-task assistive driving benchmark; real-world generalization is not demonstrated.
Reliance on frozen vision-language features may limit adaptability to novel scenarios not covered by pre-trained representations.

Resources

Paper