dVLM-AD: Enhance Diffusion Vision-Language-Model for Driving via Controllable Reasoning

TL;DR

This paper introduces dVLM-AD, a diffusion-based vision-language model for autonomous driving that improves reasoning consistency and controllability over traditional autoregressive models, enhancing safety and reliability.

Contribution

The paper proposes a novel diffusion-based VLM for driving that unifies perception, reasoning, and planning, outperforming autoregressive models in consistency and controllability.

Findings

9% improvement in behavior-trajectory consistency

6% increase in RFS on long-tail scenarios

Comparable planning performance with a modest backbone

Abstract

The autonomous driving community is increasingly focused on addressing the challenges posed by out-of-distribution (OOD) driving scenarios. A dominant research trend seeks to enhance end-to-end (E2E) driving systems by integrating vision-language models (VLMs), leveraging their rich world knowledge and reasoning abilities to improve generalization across diverse environments. However, most existing VLMs or vision-language agents (VLAs) are built upon autoregressive (AR) models. In this paper, we observe that existing AR-based VLMs -- limited by causal attention and sequential token generation -- often fail to maintain consistency and controllability between high-level reasoning and low-level planning. In contrast, recent discrete diffusion VLMs equipped with bidirectional attention exhibit superior controllability and reliability through iterative denoising. Building on these observations, we introduce dVLM-AD, a diffusion-based vision-language model that unifies perception, structured reasoning, and low-level planning for end-to-end driving. Evaluated on nuScenes and WOD-E2E, dVLM-AD yields more consistent reasoning-action pairs and achieves planning performance comparable to existing driving VLM/VLA systems despite a modest backbone, outperforming AR-based baselines with a 9 percent improvement in behavior-trajectory consistency and a 6 percent increase in RFS on long-tail WOD-E2E scenarios. These results suggest a controllable and reliable pathway for scalable end-to-end driving.