Risk-Controllable Multi-View Diffusion for Driving Scenario Generation

TL;DR

This paper introduces RiskMV-DPO, a novel diffusion-based framework for generating diverse, risk-controllable multi-view driving scenarios that enhance autonomous vehicle safety testing by synthesizing high-stakes situations with geometric consistency.

Contribution

The paper proposes a physically-informed, risk-controllable scenario generation pipeline integrating target risk levels with diffusion models, ensuring geometric fidelity and diversity in long-tail driving scenarios.

Findings

Improved 3D detection mAP from 18.17 to 30.50

Reduced FID score to 15.70

Generated diverse long-tail scenarios with high visual quality

Abstract

Generating safety-critical driving scenarios is crucial for evaluating and improving autonomous driving systems, but long-tail risky situations are rarely observed in real-world data and difficult to specify through manual scenario design. Existing generative approaches typically treat risk as an after-the-fact label and struggle to maintain geometric consistency in multi-view driving scenes. We present RiskMV-DPO, a general and systematic pipeline for physically-informed, risk-controllable multi-view scenario generation. By integrating target risk levels with physically-grounded risk modeling, we autonomously synthesize diverse and high-stakes dynamic trajectories that serve as explicit geometric anchors for a diffusion-based video generator. To ensure spatial-temporal coherence and geometric fidelity, we introduce a geometry-appearance alignment module and a region-aware direct preference optimization (RA-DPO) strategy with motion-aware masking to focus learning on localized dynamic regions.Experiments on the nuScenes dataset show that RiskMV-DPO can freely generate a wide spectrum of diverse long-tail scenarios while maintaining state-of-the-art visual quality, improving 3D detection mAP from 18.17 to 30.50 and reducing FID to 15.70. Our work shifts the role of world models from passive environment prediction to proactive, risk-controllable synthesis, providing a scalable toolchain for the safety-oriented development of embodied intelligence.