Rolling Ahead Diffusion for Traffic Scene Simulation

2025-02-13

AI Summary

Existing traffic scene simulation methods struggle to balance responsiveness and computational efficiency: diffusion models offer flexibility but incur high per-frame regeneration costs, while autoregressive models are computationally efficient yet lack long-horizon planning capability. This paper proposes Rolling Diffusion, a novel rolling-horizon diffusion generative framework featuring an innovative “ahead diffusion” mechanism. In a single inference step, it jointly outputs the next-timestep deterministic trajectory and multiple subsequent noisy future states, thereby unifying Model Predictive Control (MPC)-style real-time responsiveness with autoregressive-style inference efficiency. Leveraging rolling time-window modeling, partial noise masking, and joint multi-agent motion modeling, Rolling Diffusion achieves 98.7% trajectory plausibility and high-fidelity real-time interaction on nuScenes simulation. It accelerates full-sequence diffusion MPC by 5.3× and significantly outperforms pure autoregressive baselines in long-horizon planning quality.

Technology Category

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

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Realistic driving simulation requires that NPCs not only mimic natural driving behaviors but also react to the behavior of other simulated agents. Recent developments in diffusion-based scenario generation focus on creating diverse and realistic traffic scenarios by jointly modelling the motion of all the agents in the scene. However, these traffic scenarios do not react when the motion of agents deviates from their modelled trajectories. For example, the ego-agent can be controlled by a stand along motion planner. To produce reactive scenarios with joint scenario models, the model must regenerate the scenario at each timestep based on new observations in a Model Predictive Control (MPC) fashion. Although reactive, this method is time-consuming, as one complete possible future for all NPCs is generated per simulation step. Alternatively, one can utilize an autoregressive model (AR) to predict only the immediate next-step future for all NPCs. Although faster, this method lacks the capability for advanced planning. We present a rolling diffusion based traffic scene generation model which mixes the benefits of both methods by predicting the next step future and simultaneously predicting partially noised further future steps at the same time. We show that such model is efficient compared to diffusion model based AR, achieving a beneficial compromise between reactivity and computational efficiency.

Problem

• Simulate reactive NPC driving behaviors
• Balance reactivity and computational efficiency
• Predict immediate and future NPC movements

Innovation

• Rolling diffusion model
• Combines AR and MPC
• Efficient reactive simulation