BridgeDrive: Diffusion Bridge Policy for Closed-Loop Trajectory Planning in Autonomous Driving

2025-09-27

AI Summary

In autonomous driving’s closed-loop trajectory planning, diffusion models face challenges in guidance under complex dynamic scenes and suffer from theoretical inconsistency due to reliance on truncated scheduling. To address these issues, this paper proposes BridgeDrive—a novel anchor-guided trajectory generation method grounded in diffusion bridge policy. BridgeDrive eliminates truncated scheduling by constructing a conditional diffusion bridge, enabling theoretically consistent multimodal trajectory generation. It integrates an ODE solver to accelerate inference while satisfying real-time constraints, and establishes an end-to-end closed-loop conditional generation framework to enhance environmental responsiveness and fine-grained trajectory control. Evaluated on the Bench2Drive benchmark, BridgeDrive achieves a significant improvement in task success rate—outperforming the previous state-of-the-art by 5 percentage points—and sets a new SOTA.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Planning, Routing, and Scheduling: Planning with Language Models
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Diffusion-based planners have shown great promise for autonomous driving due to their ability to capture multi-modal driving behaviors. However, guiding these models effectively in reactive, closed-loop environments remains a significant challenge. Simple conditioning often fails to provide sufficient guidance in complex and dynamic driving scenarios. Recent work attempts to use typical expert driving behaviors (i.e., anchors) to guide diffusion models but relies on a truncated schedule, which introduces theoretical inconsistencies and can compromise performance. To address this, we introduce BridgeDrive, a novel anchor-guided diffusion bridge policy for closed-loop trajectory planning. Our approach provides a principled diffusion framework that effectively translates anchors into fine-grained trajectory plans, appropriately responding to varying traffic conditions. Our planner is compatible with efficient ODE solvers, a critical factor for real-time autonomous driving deployment. We achieve state-of-the-art performance on the Bench2Drive benchmark, improving the success rate by 5% over prior arts.

Problem

• Guiding diffusion models effectively in reactive closed-loop autonomous driving environments
• Overcoming theoretical inconsistencies from truncated schedules in anchor-guided diffusion planning
• Translating expert driving anchors into fine-grained trajectory plans for varying traffic

Innovation

• Anchor-guided diffusion bridge policy for planning
• Principled framework translating anchors to trajectories
• Compatible with efficient ODE solvers for deployment