Causal Composition Diffusion Model for Closed-loop Traffic Generation

2024-12-23

arXiv.org

AI Summary

To address the challenge of generating long-tail traffic scenarios for autonomous driving safety evaluation—where controllability and realism are difficult to jointly achieve—this paper proposes a structure-guided causal diffusion model. Methodologically, it is the first to explicitly inject causal structure into the diffusion process: traffic variables’ structural constraints are derived via causal discovery, and trajectory generation is refined through closed-loop simulation feedback. The framework unifies controllability and realism as a causal-constrained optimization problem, enabling user-intent-driven synthesis of realistic, interactive traffic flows. Evaluated on benchmark datasets and CARLA simulations, our method significantly reduces collision rate (−32.7%) and off-road rate (−41.5%), while improving final displacement error (FDE, +28.3%) and ride comfort—outperforming all existing state-of-the-art approaches comprehensively.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Search and Optimization: Sampling/Simulation-based Search
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Simulation is critical for safety evaluation in autonomous driving, particularly in capturing complex interactive behaviors. However, generating realistic and controllable traffic scenarios in long-tail situations remains a significant challenge. Existing generative models suffer from the conflicting objective between user-defined controllability and realism constraints, which is amplified in safety-critical contexts. In this work, we introduce the Causal Compositional Diffusion Model (CCDiff), a structure-guided diffusion framework to address these challenges. We first formulate the learning of controllable and realistic closed-loop simulation as a constrained optimization problem. Then, CCDiff maximizes controllability while adhering to realism by automatically identifying and injecting causal structures directly into the diffusion process, providing structured guidance to enhance both realism and controllability. Through rigorous evaluations on benchmark datasets and in a closed-loop simulator, CCDiff demonstrates substantial gains over state-of-the-art approaches in generating realistic and user-preferred trajectories. Our results show CCDiff's effectiveness in extracting and leveraging causal structures, showing improved closed-loop performance based on key metrics such as collision rate, off-road rate, FDE, and comfort.

Problem

• Generating realistic traffic scenarios
• Balancing controllability and realism
• Enhancing closed-loop simulation performance

Innovation

• Causal Compositional Diffusion Model
• Structured guidance in diffusion
• Enhanced realism and controllability