EUVIMEDEUVIMED entdecken
Crossref

Model Predictive Path Integral Control as Preconditioned Gradient Descent

Mahyar Fazlyab, Sina Sharifi, Jiarui Wang

IEEE Control Systems Letters · 2026 · 10 · 2089-2094

DOI: 10.1109/lcsys.2026.3708745 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Control Systems Letters, 2026. Verfasst von Mahyar Fazlyab und 2 weiteren Autor:innen. Die Arbeit nennt 18 Literaturangaben. Gefördert von: National Science Foundation, Internal Funds from Johns Hopkins University.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Model Predictive Path Integral (MPPI) control is a widely used sampling-based method for trajectory optimization, yet its convergence properties remain only partially understood. This letter provides a direct convergence analysis using variational optimization. By lifting constrained trajectory optimization to a Kullback-Leibler (KL) regularized problem over decision distributions, we derive a reduced free-energy objective defined over a parametric sampling family. For general parametric families, we derive gradient and Hessian representations of this reduced objective and analyze preconditioned gradient descent on the sampling-distribution parameters. In the fixed-covariance Gaussian case, the classical MPPI update is recovered exactly as a unit-step preconditioned gradient update. We prove descent and stationarity guarantees for the exact expectation-based iteration when the Hessian of the reduced objective is bounded in the metric induced by the preconditioner. For the Gaussian family, we further show that the preconditioned Hessian is governed by the covariance of the Gibbs-tilted distribution relative to the covariance of the sampling distribution, yielding a covariance-dependent sufficient condition for the descent of exact unit-step MPPI. Numerical experiments illustrate the theory and the effect of key hyperparameters.

Förderung

  • National Science Foundation
  • Internal Funds from Johns Hopkins University

Zitieren

APA

Fazlyab, M., Sharifi, S., & Wang, J. (2026). Model Predictive Path Integral Control as Preconditioned Gradient Descent. IEEE Control Systems Letters, 10, 2089-2094. https://doi.org/10.1109/lcsys.2026.3708745

Vancouver

Fazlyab M, Sharifi S, Wang J. Model Predictive Path Integral Control as Preconditioned Gradient Descent. IEEE Control Systems Letters. 2026;10:2089-2094. doi:10.1109/lcsys.2026.3708745

BibTeX
@article{Fazlyab2026,
  author = {Fazlyab, Mahyar and Sharifi, Sina and Wang, Jiarui},
  title = {Model Predictive Path Integral Control as Preconditioned Gradient Descent},
  journal = {IEEE Control Systems Letters},
  year = {2026},
  volume = {10},
  pages = {2089--2094},
  doi = {10.1109/lcsys.2026.3708745}
}
Kostenlos starten →