Search papers, labs, and topics across Lattice.
This paper introduces E2E-CDiff, an end-to-end conditional diffusion framework designed for generating realistic and controllable traffic scenarios essential for testing autonomous driving systems. By jointly denoising future motion states and low-level controls based on front-view observations, E2E-CDiff addresses the limitations of traditional two-stage planning-control methods, achieving a superior balance between controllability and realism. Experimental results on Bench2Drive indicate that E2E-CDiff not only outperforms existing reinforcement and imitation learning approaches but also effectively facilitates complex interactions for various autonomous driving systems.
E2E-CDiff achieves a groundbreaking balance between controllability and realism in traffic scenario generation, crucial for autonomous vehicle testing.
Generating closed-loop traffic scenarios that are both realistic and controllable is crucial for evaluating autonomous driving systems, especially under rare safety-critical interactions. Existing learning-based methods often struggle to balance controllability and realism, offering either limited fine-grained control over traffic behavior or controllable scenarios at the expense of behavioral plausibility. This paper presents E2E-CDiff, an end-to-end conditional diffusion framework for controllable and realistic scenario generation. Conditioned on front-view visual observations, E2E-CDiff jointly denoises future motion states and executable low-level controls for route-interacting background vehicles. This unified state-action generation mitigates the planning-control mismatch in conventional two-stage trajectory-then-controller pipelines. Differentiable guidance further regulates speed, enforces drivable-area compliance, and supports collision-avoidance or collision-seeking behaviors, enabling both naturalistic and safety-critical scenario generation. Experiments on Bench2Drive show that E2E-CDiff achieves a favorable controllability-realism trade-off compared with representative reinforcement- and imitation-learning baselines, while its collision-guided variant induces challenging interactions across multiple autonomous driving systems. E2E-CDiff also performs competitively as a learning-based ego planner, demonstrating the generality of end-to-end state-action diffusion.
