Safe and Stylized Trajectory Planning for Autonomous Driving via Diffusion Model

TL;DR

The paper tackles safe, personalized trajectory planning for autonomous driving by casting it as conditional diffusion, producing diverse but safe trajectories conditioned on a stylized context. It introduces a Multi-Source Style-Aware Encoder to extract scene- and style-relevant features and a Style-Guided Dynamic Trajectory Generator that uses time-step adaptive energy guidance to balance safety and driving style during denoising. The approach achieves state-of-the-art results on StyleDrive and NuPlan benchmarks, including first-place scores and reduced collision rates, and is validated in real-vehicle deployments on NVIDIA platforms. This work advances practical deployment by providing dynamic safety–style reconciliation, interpretability through explicit energy functions, and real-time feasibility for urban driving. The combination of dynamic attention, energy-guided diffusion, and multi-objective fusion offers a robust framework for personalized yet safe autonomous driving in complex multi-agent environments.

Abstract

Achieving safe and stylized trajectory planning in complex real-world scenarios remains a critical challenge for autonomous driving systems. This paper proposes the SDD Planner, a diffusion-based framework designed to effectively reconcile safety constraints with driving styles in real time. The framework integrates two core modules: a Multi-Source Style-Aware Encoder, which employs distance-sensitive attention to fuse dynamic agent data and environmental contexts for heterogeneous safety-style perception; and a Style-Guided Dynamic Trajectory Generator, which adaptively modulates priority weights within the diffusion denoising process to generate user-preferred yet safe trajectories. Extensive experiments demonstrate that SDD Planner achieves state-of-the-art performance. On the StyleDrive benchmark, it improves the SM-PDMS metric by 3.9% over WoTE, the strongest baseline. Furthermore, on the NuPlan Test14 and Test14-hard benchmarks, SDD Planner ranks first with overall scores of 91.76 and 80.32, respectively, outperforming leading methods such as PLUTO. Real-vehicle closed-loop tests further confirm that SDD Planner maintains high safety standards while aligning with preset driving styles, validating its practical applicability for real-world deployment.

Figures (9)

• Figure 1: Comparison of different trajectory planning paradigms. Unlike existing methods that compromise between safety and style, the proposed SDD Planner achieves user-preferred, multi-style trajectories with adaptive safety performance, approaching the ideal balance point.
• Figure 2: Overall architecture of the SDD Planner. (a) The framework encodes multi-source scenario inputs via a Multi-Source Style-Aware Encoder into high-dimensional stylized features ($\boldsymbol{z}_{style}$), serving as critical conditioning signals. (b) Style-Guided Dynamic Trajectory Generator: The diffusion decoder iteratively denoises trajectories from Gaussian noise ($t=T$) to stylized paths ($t=0$) via a time-step adaptive classifier-guided framework. This process is governed by: Style Conditioning, where $\boldsymbol{z}_{style}$ is injected via Scale Shift to enforce context consistency; and Dynamic Guidance, which modifies the score function via fused energy gradients to strictly enforce safety and style constraints.
• Figure 3: The performance of SDD Planner and PLUTO in the same scenario.
• Figure 4: Weight change chart for two different driving scenarios.
• Figure 5: Comparison of generated trajectories under different driving styles.