2022
DOI: 10.48550/arxiv.2203.01447
|View full text |Cite
Preprint
|
Sign up to set email alerts
|

Learning Stochastic Parametric Differentiable Predictive Control Policies

Abstract: The problem of synthesizing stochastic explicit model predictive control policies is known to be quickly intractable even for systems of modest complexity when using classical controltheoretic methods. To address this challenge, we present a scalable alternative called stochastic parametric differentiable predictive control (SP-DPC) for unsupervised learning of neural control policies governing stochastic linear systems subject to nonlinear chance constraints. SP-DPC is formulated as a deterministic approximat… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1

Citation Types

0
2
0

Year Published

2022
2022
2023
2023

Publication Types

Select...
1
1

Relationship

1
1

Authors

Journals

citations
Cited by 2 publications
(2 citation statements)
references
References 28 publications
0
2
0
Order By: Relevance
“…We fit model parameters to the data by minimizing the loss ( 14) with α = 0.1 using gradient descent. Model solutions were constructed in Neuromancer, an extensible library built on pytorch for differentiable parametric programming [39]. To compute the linear solutions necessary for the implicit method we used the torch.linalg.solve function as part of the pytorch package a differentiable routine that allows for the back-propagation of gradients for optimization [40].…”
Section: Methodsmentioning
confidence: 99%
“…We fit model parameters to the data by minimizing the loss ( 14) with α = 0.1 using gradient descent. Model solutions were constructed in Neuromancer, an extensible library built on pytorch for differentiable parametric programming [39]. To compute the linear solutions necessary for the implicit method we used the torch.linalg.solve function as part of the pytorch package a differentiable routine that allows for the back-propagation of gradients for optimization [40].…”
Section: Methodsmentioning
confidence: 99%
“…DPC brings forth the idea of offline computation of explicit predictive control policies by leveraging automatic differentiation (AD) of the constrained optimization problem for direct computation of the policy gradients. In our previous work, we have demonstrated the scalability of the DPC framework on systems, including uncertainties and nonlinear constraints [32], [33]. Contributions.…”
Section: Introductionmentioning
confidence: 99%