← Publications
arXiv2026Preprint

Driving Intents Amplify Planning-Oriented Reinforcement Learning

Hengtong Lu, Victor Shea-Jay Huang, Chengmin Yang, Pengfei Jing, Jifeng Dai, Yan Xie, Benjin Zhu

arXiv Project Citation
DIAL contrasts collapsed trajectory samples with diverse intent-conditioned proposals and plots their best-of-K preference scores.
DIAL contrasts collapsed trajectory samples with diverse intent-conditioned proposals and plots their best-of-K preference scores.

TL;DR

Intent-conditioned proposals and multi-intent GRPO expand and preserve diverse driving maneuvers for preference-based reinforcement learning.通过意图条件采样与多意图 GRPO，扩展并保留多样驾驶行为，为基于偏好的强化学习提供更丰富的候选轨迹。

Citation

Download .bib
@article{lu2026driving,
  title = {Driving Intents Amplify Planning-Oriented Reinforcement Learning},
  author = {Lu, Hengtong and Huang, Victor Shea-Jay and Yang, Chengmin and Jing, Pengfei and Dai, Jifeng and Xie, Yan and Zhu, Benjin},
  journal = {arXiv preprint arXiv:2605.12625},
  year = {2026},
  url = {https://arxiv.org/abs/2605.12625}
}