Journal article · 2026 · Open access

Enhancing autonomous vehicle safety: A hybrid MPC-DRL-SMC approach validated via CARLA and historical crash records

Teshome Kumsa Kurse, Girma Gebresenbet, Geleta Fikadu Daba, Negasa Abdena Alemu

Results in Engineering, 30, 109912 · Published 10 March 2026 · DOI 10.1016/j.rineng.2026.109912

Open full textCite this paper

In brief

“Enhancing autonomous vehicle safety: A hybrid MPC-DRL-SMC approach validated via CARLA and historical crash records” is a 2026 journal article by Teshome Kumsa Kurse and 3 co-authors, published in Results in Engineering. OpenAlex counts 1 citations as of 22 November 2026. It is open access under a CC BY-NC-ND licence: the full text is free to read at Results in Engineering.

Citations
1
Most cited in
2026 (1)
Field-weighted impact
1.09
References
36
Authors
4

Abstract

Ensuring autonomous vehicle (AV) safety in rare, complex edge cases remains a critical barrier to real-world deployment, primarily due to the gap between statistical simulation performance and empirical robustness. This paper introduces a novel triple-layer hybrid control architecture that integrates Model Predictive Control (MPC), Deep Reinforcement Learning (DRL), and Sliding Mode Control (SMC). The framework is uniquely grounded in 9.6 years of real-world crash data from the California DMV (2014–2023), from which a data-driven risk model was calibrated against 628 historical collision scenarios. This empirical foundation enables a novel coherent risk measure to dynamically adjust the controller’s safety priority via a DRL policy that modulates MPC cost weight in real time. We establish five foundational theorems providing formal guarantees for system convergence, Pareto optimality, and bounded tracking error. Validated through high-fidelity CARLA simulations, the proposed MPC-DRL-SMC framework achieves a 78% improvement in safety risk scores, a 71% reduction in tracking error, and a 35% reduction in control effort compared to standard MPC while maintaining real-time operation at 16.1 Hz. The core methodological novelty lies in the real-time, bounded coupling of a Deep Reinforcement Learning policy to a Model Predictive Controller via an adaptive weight mechanism guided by a crash calibrated risk measure and backed by formal stability guarantees. This work delivers a uniform framework that simultaneously achieves data-driven risk-awareness, mathematical safety certification, and comprehensive real-world validation a combination not present in prior hybrid MPC-RL architectures.

Cite this paper

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

APA 7

Kurse, T. K., Gebresenbet, G., Daba, G. F., & Alemu, N. A. (2026). Enhancing autonomous vehicle safety: A hybrid MPC-DRL-SMC approach validated via CARLA and historical crash records. Results in Engineering, 30, 109912. https://doi.org/10.1016/j.rineng.2026.109912

MLA 9

Kurse, Teshome Kumsa, et al. “Enhancing autonomous vehicle safety: A hybrid MPC-DRL-SMC approach validated via CARLA and historical crash records.” Results in Engineering, vol. 30, 2026, p. 109912. https://doi.org/10.1016/j.rineng.2026.109912.

Chicago

Kurse, Teshome Kumsa, Girma Gebresenbet, Geleta Fikadu Daba, and Negasa Abdena Alemu. 2026. “Enhancing autonomous vehicle safety: A hybrid MPC-DRL-SMC approach validated via CARLA and historical crash records.” Results in Engineering 30: 109912. https://doi.org/10.1016/j.rineng.2026.109912.

Harvard

Kurse, T. K. et al. (2026) ‘Enhancing autonomous vehicle safety: A hybrid MPC-DRL-SMC approach validated via CARLA and historical crash records’, Results in Engineering, 30, p. 109912. doi:10.1016/j.rineng.2026.109912.

IEEE

T. K. Kurse, G. Gebresenbet, G. F. Daba, and N. A. Alemu, “Enhancing autonomous vehicle safety: A hybrid MPC-DRL-SMC approach validated via CARLA and historical crash records,” Results in Engineering, vol. 30, p. 109912, 2026, doi: 10.1016/j.rineng.2026.109912.

BibTeX

@article{kurse2026enhancing,
  title     = {{Enhancing autonomous vehicle safety: A hybrid MPC-DRL-SMC approach validated via CARLA and historical crash records}},
  author    = {Kurse, Teshome Kumsa and Gebresenbet, Girma and Daba, Geleta Fikadu and Alemu, Negasa Abdena},
  journal   = {Results in Engineering},
  year      = {2026},
  volume    = {30},
  pages     = {109912},
  doi       = {10.1016/j.rineng.2026.109912},
  url       = {https://doi.org/10.1016/j.rineng.2026.109912}
}

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

Topic and keywords

Physical Sciences › Engineering › Automotive Engineering › Vehicle Dynamics and Control Systems

  • autonomous vehicle safety
  • model predictive control
  • deep reinforcement learning
  • sliding mode control
  • stability guarantees
  • Pareto optimality
  • tracking error
  • CARLA simulator

Work with this paper

Most-cited papers in Engineering

All highly cited Engineering papers

Questions about this paper

How do I cite this paper in APA style?

Kurse, T. K., Gebresenbet, G., Daba, G. F., & Alemu, N. A. (2026). Enhancing autonomous vehicle safety: A hybrid MPC-DRL-SMC approach validated via CARLA and historical crash records. Results in Engineering, 30, 109912. https://doi.org/10.1016/j.rineng.2026.109912

How many times has this paper been cited?

1 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 Results in Engineering: sign in and open the full text from this page.

Who wrote this paper?

Teshome Kumsa Kurse, Girma Gebresenbet, Geleta Fikadu Daba, Negasa Abdena Alemu.

What is the DOI of this paper?

10.1016/j.rineng.2026.109912, permanently at https://doi.org/10.1016/j.rineng.2026.109912. 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.