Awesome World Model Hub Papers · Datasets · Projects
← Back to papers

Imagine-2-Drive: High-Fidelity World Modeling in CARLA for Autonomous Vehicles

arXiv 24.11 2024 66 method, application

TLDR

Proposes Imagine-2-Drive, a WMRL framework with diffusion-based world model and policy for autonomous driving, outperforming baselines in CARLA.

Reasoning

Strengths include novel integration of diffusion models for high-fidelity world modeling and multi-modal policy, addressing error accumulation and decision-making diversity. Weaknesses are evaluation limited to the CARLA simulator without real-world validation, and reliance on simulated benchmarks.

Read-first score

Read-first score 66, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 55.

Topical relevance 42%
78.6

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

Recency 8%
75.1

Uses a gentle age decay so recent papers surface without erasing older foundations. 2024

Methodology quality 25%
70

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

Reproducibility 25%
38

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

Field roles

Methodology anchor

Rank sensitivity

Stability: volatile; rank range: 229.

Keyword Scores

world model
10
model-based reinforcement learning world model
10
generative world model
9
world dynamics prediction
8
video world model
7
interactive world model
6
world simulator
5

Deep Analysis

Innovations

  • DiffDreamer: a diffusion-based world model that generates future observations simultaneously to mitigate error accumulation
  • DPA (Diffusion Policy Actor): a diffusion-based policy that models diverse and multi-modal trajectory distributions
  • Integration of diffusion-based world model and policy actor into a WMRL framework for sample-efficient autonomous driving

Methodology

Imagine-2-Drive is a World Model-based Reinforcement Learning (WMRL) framework combining a diffusion-based world model (DiffDreamer) that generates future observations in parallel to reduce compounding errors, and a diffusion policy actor (DPA) that captures multi-modal trajectory distributions. The policy is trained within the world model using minimal online interactions, and evaluation is conducted in the CARLA simulator using standard driving benchmarks.

Key Results

The method outperforms prior world model baselines, achieving a 15% improvement in Route Completion and a 20% improvement in Success Rate.

Tags