Conference paper · 2023 · Open access

Evaluating Model-Free Reinforcement Learning toward Safety-Critical Tasks

Linrui Zhang, Qin Zhang, Shen Li, Bo Yuan, Xueqian Wang, Dacheng Tao

Proceedings of the AAAI Conference on Artificial Intelligence, 37(12), 15313–15321 · Published 26 June 2023 · DOI 10.1609/aaai.v37i12.26786

Open PDFView at the publisherCite this paper

In brief

“Evaluating Model-Free Reinforcement Learning toward Safety-Critical Tasks” is a 2023 conference paper by Linrui Zhang and 5 co-authors, published in Proceedings of the AAAI Conference on Artificial Intelligence. OpenAlex counts 55 citations as of 22 November 2026, which puts it in the top 4% of works published in 2023 by citations. It is open access: the full text is free to read at Proceedings of the AAAI Conference on Artificial Intelligence.

Citations
55
Most cited in
2024 (20)
Field-weighted impact
3.43
References
36
Authors
6

Abstract

Safety comes first in many real-world applications involving autonomous agents. Despite a large number of reinforcement learning (RL) methods focusing on safety-critical tasks, there is still a lack of high-quality evaluation of those algorithms that adheres to safety constraints at each decision step under complex and unknown dynamics. In this paper, we revisit prior work in this scope from the perspective of state-wise safe RL and categorize them as projection-based, recovery-based, and optimization-based approaches, respectively. Furthermore, we propose Unrolling Safety Layer (USL), a joint method that combines safety optimization and safety projection. This novel technique explicitly enforces hard constraints via the deep unrolling architecture and enjoys structural advantages in navigating the trade-off between reward improvement and constraint satisfaction. To facilitate further research in this area, we reproduce related algorithms in a unified pipeline and incorporate them into SafeRL-Kit, a toolkit that provides off-the-shelf interfaces and evaluation utilities for safety-critical tasks. We then perform a comparative study of the involved algorithms on six benchmarks ranging from robotic control to autonomous driving. The empirical results provide an insight into their applicability and robustness in learning zero-cost-return policies without task-dependent handcrafting. The project page is available at https://sites.google.com/view/saferlkit.

Cite this paper

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

APA 7

Zhang, L., Zhang, Q., Li, S., Yuan, B., Wang, X., & Tao, D. (2023). Evaluating Model-Free Reinforcement Learning toward Safety-Critical Tasks. Proceedings of the AAAI Conference on Artificial Intelligence, 37(12), 15313–15321. https://doi.org/10.1609/aaai.v37i12.26786

MLA 9

Zhang, Linrui, et al. “Evaluating Model-Free Reinforcement Learning toward Safety-Critical Tasks.” Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, no. 12, 2023, pp. 15313–15321. https://doi.org/10.1609/aaai.v37i12.26786.

Chicago

Zhang, Linrui, Qin Zhang, Shen Li, Bo Yuan, Xueqian Wang, and Dacheng Tao. 2023. “Evaluating Model-Free Reinforcement Learning toward Safety-Critical Tasks.” Proceedings of the AAAI Conference on Artificial Intelligence 37 (12): 15313–15321. https://doi.org/10.1609/aaai.v37i12.26786.

Harvard

Zhang, L. et al. (2023) ‘Evaluating Model-Free Reinforcement Learning toward Safety-Critical Tasks’, Proceedings of the AAAI Conference on Artificial Intelligence, 37(12), pp. 15313–15321. doi:10.1609/aaai.v37i12.26786.

IEEE

L. Zhang, Q. Zhang, S. Li, B. Yuan, X. Wang, and D. Tao, “Evaluating Model-Free Reinforcement Learning toward Safety-Critical Tasks,” Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, no. 12, pp. 15313–15321, 2023, doi: 10.1609/aaai.v37i12.26786.

BibTeX

@inproceedings{zhang2023evaluating,
  title     = {{Evaluating Model-Free Reinforcement Learning toward Safety-Critical Tasks}},
  author    = {Zhang, Linrui and Zhang, Qin and Li, Shen and Yuan, Bo and Wang, Xueqian and Tao, Dacheng},
  booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year      = {2023},
  volume    = {37},
  number    = {12},
  pages     = {15313--15321},
  doi       = {10.1609/aaai.v37i12.26786},
  url       = {https://doi.org/10.1609/aaai.v37i12.26786}
}

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

Citations by year

YearCitationsShare of the best year
202344
20242020
20251717
20261414

Its field-weighted citation impact is 3.43: it has been cited 3.43 times as often as the average work of the same field, type and year (1.00 is the world average).

Topic and keywords

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

  • model-free reinforcement learning
  • safety constraints
  • robot control
  • autonomous driving

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?

Zhang, L., Zhang, Q., Li, S., Yuan, B., Wang, X., & Tao, D. (2023). Evaluating Model-Free Reinforcement Learning toward Safety-Critical Tasks. Proceedings of the AAAI Conference on Artificial Intelligence, 37(12), 15313–15321. https://doi.org/10.1609/aaai.v37i12.26786

How many times has this paper been cited?

55 times, according to OpenAlex on 22 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 Proceedings of the AAAI Conference on Artificial Intelligence: sign in and open the PDF from this page.

Who wrote this paper?

Linrui Zhang, Qin Zhang, Shen Li, Bo Yuan, Xueqian Wang, Dacheng Tao.

What is the DOI of this paper?

10.1609/aaai.v37i12.26786, permanently at https://doi.org/10.1609/aaai.v37i12.26786. Sign in and open it from this page.

Bibliographic data from OpenAlex (CC0), last checked 22 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.