KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#dsh #dsh-plugin #codex #typescript #developer-tools #ai-agents #deepseek-harness #claude-code #ai-agent #cli #react #mcp

FlowDrive

★ 54 Open GitHub ↗

FlowDrive: Energy Flow Field for End-to-End Autonomous Driving

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

<div align="center">
<h3>FlowDrive: Energy Flow Field for End-to-End Autonomous Driving</h3>

[Hao Jiang](https://github.com/Sunstroperao)1, [Zhipeng Zhang](https://scholar.google.com/citations?hl=zh-CN&user=7Ws0QHYAAAAJ)1, Yu Gao2, [Zhigang Sun](https://github.com/SunZhigang7)2, Yiru Wang2, Yuwen Heng2, Shuo Wang2, Jinhao Chai4, Zhuo Chen1, Hao Zhao3, Hao Sun2, Xi Zhang1, [Anqing Jiang](https://github.com/PeterJaq)2✉, [Chuan Hu](https://scholar.google.com/citations?hl=zh-CN&user=xRgru9YAAAAJ)1✉

1Shanghai Jiao Tong University 
2Bosch Corporate Research, Shanghai, China
3AIR, Tsinghua University
4Shanghai University

(✉) Corresponding author.

</div>

![visualization_example](https://raw.githubusercontent.com/AstrixDrive/FlowDrive/main/assets/example.gif)

## News

* **`[2025/09/19]`:** We released our paper on [Arxiv](https://arxiv.org/abs/2509.14303). Code/Models are coming soon. Please stay tuned! ☕️

## Updates
We are going to release code step by step:
- [x] Release Paper  
- [ ] Release Full Models and Training/Evaluation Framework   
- [ ] Release updated paper

**Please note:** The codebase is currently undergoing refactoring and polishing to prepare for its public release. We are committed to open-sourcing the full project and plan to do so in the coming months.

## Table of Contents
- [Abstract](#Abstract)
- [Getting Started](#getting-started)
- [Acknowledgement](#acknowledgement)
- [Citation](#citation)

## Abstract 
<div align="justify">
Recent advances in end-to-end autonomous driving leverage multi-view images to construct BEV representations for motion planning. In motion planning, autonomous vehicles need considering both hard constraints imposed by geometrically occupied obstacles (e.g., vehicles, pedestrians) and soft, rule-based semantics with no explicit geometry (e.g., lane boundaries, traffic priors). However, existing end-to-end frameworks typically rely on BEV features learned in an implicit manner, lacking explicit modeling of risk and guidance priors for safe and interpretable planning. To address this, we propose FlowDrive, a novel framework that introduces physically interpretable energy-based flow fields—including risk potential and lane attraction fields—to encode semantic priors and safety cues into the BEV space. These flow-aware features enable adaptive refinement of anchor trajectories and serve as interpretable guidance for trajectory generation. Moreover, FlowDrive decouples motion intent prediction from trajectory denoising via a conditional diffusion planner with feature-level gating, alleviating task interference and enhancing multimodal diversity. Experiments on the NAVSIM v2 benchmark demonstrate that FlowDrive achieves state-of-the-art performance with an EPDMS of 86.3, surpassing prior baselines in both safety and planning quality.

![framework](https://raw.githubusercontent.com/AstrixDrive/FlowDrive/main/assets/framework.png)
</div>

## Getting Started
- [Download NAVSIM datasets following official instruction](https://github.com/autonomousvision/navsim/blob/main/docs/install.md)
- [Preparation of FlowDrive environment](docs/install.md)
- [FlowDrive Training and Evaluation](docs/train_eval.md)

## Acknowledgement

FlowDrive is greatly inspired by the following outstanding contributions to the open-source community: [NAVSIM](https://github.com/autonomousvision/navsim), [Transfuser](https://github.com/autonomousvision/transfuser), [DiffusionDrive](https://github.com/hustvl/DiffusionDrive), [GoalFlow](https://github.com/YvanYin/GoalFlow).

## Citation

If you find FlowDrive is useful in your research or applications, please consider giving us a star 🌟 and citing it by the following BibTeX entry.

```bibtex
@misc{jiang2025flowdriveenergyflowfield,
      title={FlowDrive: Energy Flow Field for End-to-End Autonomous Driving}, 
      author={Hao Jiang and Zhipeng Zhang and Yu Gao and Zhigang Sun and Yiru Wang and Yuwen Heng and Shuo Wang and Jinhao Chai and Zhuo Chen and Hao Zhao and Hao Sun and Xi Zhang and Anqing Jiang and Chuan Hu},
      year={2025},
      eprint={2509.14303},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2509.14303}, 
}
```

FlowDrive

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README