FlowDrive: moderated flow matching with data balancing for trajectory planning

2025-09-26

AI Summary

In autonomous driving trajectory planning, learning-based models suffer from poor generalization to rare hazardous scenarios due to the long-tailed distribution of real-world driving data. To address this, we propose a flow-matching-based generative trajectory planning framework. First, we introduce a trajectory-pattern-aware reweighting strategy to mitigate class imbalance in training data. Second, we design a cyclic flow matching architecture with moderate guidance to enhance trajectory diversity while preserving scene consistency. Third, we adopt a conditional denoising flow model, leveraging few matching steps and a cyclic perturbation mechanism for efficient, lightweight trajectory distribution modeling. Evaluated on nuPlan and interPlan benchmarks, our method achieves state-of-the-art performance among learning-based approaches. With a lightweight post-processing module, it matches or even surpasses certain rule-augmented methods in overall performance.

Technology Category

• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Multiagent Systems: Multiagent Planning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Learning-based planners are sensitive to the long-tailed distribution of driving data. Common maneuvers dominate datasets, while dangerous or rare scenarios are sparse. This imbalance can bias models toward the frequent cases and degrade performance on critical scenarios. To tackle this problem, we compare balancing strategies for sampling training data and find reweighting by trajectory pattern an effective approach. We then present FlowDrive, a flow-matching trajectory planner that learns a conditional rectified flow to map noise directly to trajectory distributions with few flow-matching steps. We further introduce moderated, in-the-loop guidance that injects small perturbation between flow steps to systematically increase trajectory diversity while remaining scene-consistent. On nuPlan and the interaction-focused interPlan benchmarks, FlowDrive achieves state-of-the-art results among learning-based planners and approaches methods with rule-based refinements. After adding moderated guidance and light post-processing (FlowDrive*), it achieves overall state-of-the-art performance across nearly all benchmark splits.

Problem

• Addresses data imbalance in learning-based trajectory planners for autonomous driving
• Improves performance on rare and dangerous driving scenarios through data balancing
• Enhances trajectory diversity while maintaining scene consistency in planning

Innovation

• Flow matching with conditional rectified flow
• Moderated guidance increasing trajectory diversity
• Data balancing by trajectory pattern reweighting