Slack-based stochastic Model Predictive Control
This article introduces a Model Predictive Control (MPC) framework that leverages a pair of complementary slack variables to quantify constraint violation and robustness, enabling linear probabilistic constraint enforcement. Central to this approach is the definition of an interpretable slack ratio that serves as a monotonic proxy for the violation probability under a mild stochastic-ordering assumption that admits decision-dependent dispersion of the constraint margin, beyond additive-disturbance settings. For location–scale distributions (e.g., Gaussian, Uniform, Laplace), this ratio is invariant to the disturbance mean and variance for a fixed violation probability. We exploit this property to pre-compute low-complexity safety margins that guarantee constraint satisfaction with high confidence at the first prediction step—and along the whole horizon for disturbance families closed under convolution, e.g., the Gaussian case—eliminating the need for online variance estimation. Simulation results demonstrate the applicability of the proposed method in practical problems.
Authors
- Takeshi Hatanaka (ORCID: https://orcid.org/0000-0003-3143-121X)
- Jose Maria Maestre (ORCID: https://orcid.org/0000-0002-4968-6811)
- Andrés Hernández-Rivera (ORCID: https://orcid.org/0000-0002-8418-1333)
Institutions
- Institute of Science Tokyo (JP)
- Universidad de Sevilla (ES)
Publication Details
- Journal
- Automatica
- Published
- 2026-11-20
- DOI
- https://doi.org/10.1016/j.automatica.2026.113363
- Primary Topic
- Advanced Control Systems Optimization
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Ministerio de Ciencia, Innovación y Universidades