ResWorld: Temporal Residual World Model for End-to-End Autonomous Driving

ICLR 26 2026 64.2 method, application

TLDR

Proposes temporal residual world model for autonomous driving, focusing on dynamic objects and trajectory refinement, achieving SOTA on nuScenes and NAVSIM.

Reasoning

Strengths include novel temporal residual approach to avoid static redundancy and trajectory refinement module. Weaknesses are domain specificity to driving and lack of broader world model generality.

Read-first score

Read-first score 64.2, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 33.

Recency 8%

100

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

Reproducibility 25%

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

Methodology quality 25%

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

Topical relevance 42%

47.1

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

Field roles

FrontierReproducibility anchor

Rank sensitivity

Stability: volatile; rank range: 436.

Keyword Scores

world model

world dynamics prediction

generative world model

interactive world model

world simulator

video world model

model-based reinforcement learning world model

Deep Analysis

Innovations

Temporal Residual World Model (TR-World) that focuses on dynamic object modeling by calculating temporal residuals of scene representations without relying on detection and tracking.
Future-Guided Trajectory Refinement (FGTR) module that interacts prior trajectories with future BEV features to refine trajectories and provides sparse spatial-temporal supervision to prevent world model collapse.

Methodology

The method uses temporal residuals of scene representations to extract dynamic object information without detection or tracking, then predicts the future spatial distribution of dynamic objects. These predictions are combined with static object information from current BEV features to obtain accurate future BEV features. A Future-Guided Trajectory Refinement module interacts prior trajectories with future BEV features to refine trajectories and provides sparse spatial-temporal supervision to prevent world model collapse.

Key Results

Comprehensive experiments on the nuScenes and NAVSIM datasets demonstrate that ResWorld achieves state-of-the-art planning performance.

Resources

Paper Code