Journal article · 2026 · Open access

RAD-2: Scaling Reinforcement Learning in a Generator-Discriminator Framework

Hao Gao, Shaoyu Chen, Yifan Zhu, Yuehao Song, Wenyu Liu, Qian W. L. Zhang, Xinggang Wang

IEEE Transactions on Pattern Analysis and Machine Intelligence, PP, 1–18 · Published 1 January 2026 · DOI 10.1109/tpami.2026.3749808

Open full textCite this paper

In brief

“RAD-2: Scaling Reinforcement Learning in a Generator-Discriminator Framework” is a 2026 journal article by Hao Gao and 6 co-authors, published in IEEE Transactions on Pattern Analysis and Machine Intelligence. OpenAlex counts 6 citations as of 16 November 2026, which puts it in the top 2% of works published in 2026 by citations. It is open access under a CC BY licence: the full text is free to read at IEEE Transactions on Pattern Analysis and Machine Intelligence.

Citations
6
Most cited in
2026 (6)
Field-weighted impact
8.37
References
44
Authors
7

Abstract

High-level autonomous driving requires motion planners capable of modeling multimodal future uncertainties while remaining robust in closed-loop interactions. Although diffusion-based planners are effective at modeling complex trajectory distributions, they often suffer from stochastic instabilities and the lack of corrective negative feedback when trained purely with imitation learning. To address these issues, we propose RAD-2, a unified generator-discriminator framework for closed-loop planning. Specifically, a diffusion-based generator is used to produce diverse trajectory candidates, while an RL-optimized discriminator reranks these candidates according to their long-term driving quality. This decoupled design avoids directly applying sparse scalar rewards to the full high-dimensional trajectory space, thereby improving optimization stability. To further enhance reinforcement learning, we introduce Temporally Consistent Group Relative Policy Optimization, which exploits temporal coherence to alleviate the credit assignment problem. In addition, we propose On-policy Generator Optimization, which converts closed-loop feedback into structured longitudinal optimization signals and progressively shifts the generator toward high-reward trajectory manifolds. To support efficient large-scale training, we introduce BEV-Warp, a high- throughput simulation environment that performs closed-loop evaluation directly in Bird's-Eye View feature space via spatial warping. RAD-2 reduces the collision rate by 56% compared with strong diffusion-based planners. Preliminary real-world mainpaper validation on four routes across two cities further indicates improved perceived safety and driving smoothness in complex urban traffic. Videos of closed-loop simulations, real-world deployments, and comparative experiments are available at https://hgao-cv.github.io/RAD-2.

Cite this paper

Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.

APA 7

Gao, H., Chen, S., Zhu, Y., Song, Y., Liu, W., Zhang, Q. W. L., & Wang, X. (2026). RAD-2: Scaling Reinforcement Learning in a Generator-Discriminator Framework. IEEE Transactions on Pattern Analysis and Machine Intelligence, PP, 1–18. https://doi.org/10.1109/tpami.2026.3749808

MLA 9

Gao, Hao, et al. “RAD-2: Scaling Reinforcement Learning in a Generator-Discriminator Framework.” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PP, 2026, pp. 1–18. https://doi.org/10.1109/tpami.2026.3749808.

Chicago

Gao, Hao, Shaoyu Chen, Yifan Zhu, Yuehao Song, Wenyu Liu, Qian W. L. Zhang, and Xinggang Wang. 2026. “RAD-2: Scaling Reinforcement Learning in a Generator-Discriminator Framework.” IEEE Transactions on Pattern Analysis and Machine Intelligence PP: 1–18. https://doi.org/10.1109/tpami.2026.3749808.

Harvard

Gao, H. et al. (2026) ‘RAD-2: Scaling Reinforcement Learning in a Generator-Discriminator Framework’, IEEE Transactions on Pattern Analysis and Machine Intelligence, PP, pp. 1–18. doi:10.1109/tpami.2026.3749808.

IEEE

H. Gao et al., “RAD-2: Scaling Reinforcement Learning in a Generator-Discriminator Framework,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PP, pp. 1–18, 2026, doi: 10.1109/tpami.2026.3749808.

BibTeX

@article{gao2026scaling,
  title     = {{RAD-2: Scaling Reinforcement Learning in a Generator-Discriminator Framework}},
  author    = {Gao, Hao and Chen, Shaoyu and Zhu, Yifan and Song, Yuehao and Liu, Wenyu and Zhang, Qian W. L. and Wang, Xinggang},
  journal   = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year      = {2026},
  volume    = {PP},
  pages     = {1--18},
  doi       = {10.1109/tpami.2026.3749808},
  url       = {https://doi.org/10.1109/tpami.2026.3749808}
}

Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)

Topic and keywords

Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics

  • autonomous driving
  • motion planning
  • closed-loop planning
  • trajectory generation
  • credit assignment
  • collision avoidance

Work with this paper

Most-cited papers in Computer Science

All highly cited Computer Science papers

Questions about this paper

How do I cite this paper in APA style?

Gao, H., Chen, S., Zhu, Y., Song, Y., Liu, W., Zhang, Q. W. L., & Wang, X. (2026). RAD-2: Scaling Reinforcement Learning in a Generator-Discriminator Framework. IEEE Transactions on Pattern Analysis and Machine Intelligence, PP, 1–18. https://doi.org/10.1109/tpami.2026.3749808

How many times has this paper been cited?

6 times, according to OpenAlex on 16 November 2026. Google Scholar and Scopus count different sets of sources, so their numbers differ.

Is this paper free to read?

Yes. It is open access at IEEE Transactions on Pattern Analysis and Machine Intelligence: sign in and open the full text from this page.

Who wrote this paper?

Hao Gao, Shaoyu Chen, Yifan Zhu, Yuehao Song, Wenyu Liu, Qian W. L. Zhang, Xinggang Wang.

What is the DOI of this paper?

10.1109/tpami.2026.3749808, permanently at https://doi.org/10.1109/tpami.2026.3749808. Sign in and open it from this page.

Bibliographic data from OpenAlex (CC0), last checked 16 November 2026. ResearchGuru does not host this paper; once you are signed in, the links open it at the publisher or the repository that does.