TakeAD: Preference-based Post-optimization for End-to-end Autonomous Driving with Expert Takeover Data
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911331301457920 |
|---|---|
| author | Liu, Deqing Gao, Yinfeng Qian, Deheng Zhang, Qichao Ye, Xiaoqing Han, Junyu Zheng, Yupeng Liu, Xueyi Xia, Zhongpu Ding, Dawei Pan, Yifeng Zhao, Dongbin |
| author_facet | Liu, Deqing Gao, Yinfeng Qian, Deheng Zhang, Qichao Ye, Xiaoqing Han, Junyu Zheng, Yupeng Liu, Xueyi Xia, Zhongpu Ding, Dawei Pan, Yifeng Zhao, Dongbin |
| contents | Existing end-to-end autonomous driving methods typically rely on imitation learning (IL) but face a key challenge: the misalignment between open-loop training and closed-loop deployment. This misalignment often triggers driver-initiated takeovers and system disengagements during closed-loop execution. How to leverage those expert takeover data from disengagement scenarios and effectively expand the IL policy's capability presents a valuable yet unexplored challenge. In this paper, we propose TakeAD, a novel preference-based post-optimization framework that fine-tunes the pre-trained IL policy with this disengagement data to enhance the closed-loop driving performance. First, we design an efficient expert takeover data collection pipeline inspired by human takeover mechanisms in real-world autonomous driving systems. Then, this post optimization framework integrates iterative Dataset Aggregation (DAgger) for imitation learning with Direct Preference Optimization (DPO) for preference alignment. The DAgger stage equips the policy with fundamental capabilities to handle disengagement states through direct imitation of expert interventions. Subsequently, the DPO stage refines the policy's behavior to better align with expert preferences in disengagement scenarios. Through multiple iterations, the policy progressively learns recovery strategies for disengagement states, thereby mitigating the open-loop gap. Experiments on the closed-loop Bench2Drive benchmark demonstrate our method's effectiveness compared with pure IL methods, with comprehensive ablations confirming the contribution of each component. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_17370 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | TakeAD: Preference-based Post-optimization for End-to-end Autonomous Driving with Expert Takeover Data Liu, Deqing Gao, Yinfeng Qian, Deheng Zhang, Qichao Ye, Xiaoqing Han, Junyu Zheng, Yupeng Liu, Xueyi Xia, Zhongpu Ding, Dawei Pan, Yifeng Zhao, Dongbin Robotics Artificial Intelligence Existing end-to-end autonomous driving methods typically rely on imitation learning (IL) but face a key challenge: the misalignment between open-loop training and closed-loop deployment. This misalignment often triggers driver-initiated takeovers and system disengagements during closed-loop execution. How to leverage those expert takeover data from disengagement scenarios and effectively expand the IL policy's capability presents a valuable yet unexplored challenge. In this paper, we propose TakeAD, a novel preference-based post-optimization framework that fine-tunes the pre-trained IL policy with this disengagement data to enhance the closed-loop driving performance. First, we design an efficient expert takeover data collection pipeline inspired by human takeover mechanisms in real-world autonomous driving systems. Then, this post optimization framework integrates iterative Dataset Aggregation (DAgger) for imitation learning with Direct Preference Optimization (DPO) for preference alignment. The DAgger stage equips the policy with fundamental capabilities to handle disengagement states through direct imitation of expert interventions. Subsequently, the DPO stage refines the policy's behavior to better align with expert preferences in disengagement scenarios. Through multiple iterations, the policy progressively learns recovery strategies for disengagement states, thereby mitigating the open-loop gap. Experiments on the closed-loop Bench2Drive benchmark demonstrate our method's effectiveness compared with pure IL methods, with comprehensive ablations confirming the contribution of each component. |
| title | TakeAD: Preference-based Post-optimization for End-to-end Autonomous Driving with Expert Takeover Data |
| topic | Robotics Artificial Intelligence |
| url | https://arxiv.org/abs/2512.17370 |