2000 character limit reached
BP-MPC: Optimizing the Closed-Loop Performance of MPC using BackPropagation (2312.15521v3)
Published 24 Dec 2023 in math.OC, cs.SY, and eess.SY
Abstract: Model predictive control (MPC) is pervasive in research and industry. However, designing the cost function and the constraints of the MPC to maximize closed-loop performance remains an open problem. To achieve optimal tuning, we propose a backpropagation scheme that solves a policy optimization problem with nonlinear system dynamics and MPC policies. We enforce the system dynamics using linearization and allow the MPC problem to contain elements that depend on the current system state and on past MPC solutions. Moreover, we propose a simple extension that can deal with losses of feasibility. Our approach, unlike other methods in the literature, enjoys convergence guarantees.
- “A quasi-infinite horizon nonlinear model predictive control scheme with guaranteed stability” In Automatica 34.10 Elsevier, 1998, pp. 1205–1217
- Stefano Di Cairano and Alberto Bemporad “Model predictive control tuning by controller matching” In IEEE Transactions on Automatic Control 55.1 IEEE, 2009, pp. 185–190
- Raffaele Soloperto, Johannes Köhler and Frank Allgöwer “Augmenting MPC schemes with active learning: Intuitive tuning and guaranteed performance” In IEEE Control Systems Letters 4.3 IEEE, 2020, pp. 713–718
- “Deterministic policy gradient algorithms” In International conference on machine learning, 2014, pp. 387–395 Pmlr
- Brandon Amos and J Zico Kolter “Optnet: Differentiable optimization as a layer in neural networks” In International Conference on Machine Learning, 2017, pp. 136–145 PMLR
- Asen L Dontchev, R Tyrrell Rockafellar and R Tyrrell Rockafellar “Implicit functions and solution mappings: A view from variational analysis” Springer, 2009
- “Differentiable mpc for end-to-end planning and control” In Advances in neural information processing systems 31, 2018
- Alex Oshin and Evangelos A Theodorou “Differentiable Robust Model Predictive Control” In arXiv preprint arXiv:2308.08426, 2023
- Ewa M Bednarczuk and Krzysztof E Rutkowski “On Lipschitz continuity of projections onto polyhedral moving sets” In Applied Mathematics & Optimization 84.2 Springer, 2021, pp. 2147–2175
- “Conservative set valued fields, automatic differentiation, stochastic gradient methods and deep learning” In Mathematical Programming 188 Springer, 2021, pp. 19–51
- “Stochastic subgradient method converges on tame functions” In Foundations of computational mathematics 20.1 Springer, 2020, pp. 119–154
- “Learning convex optimization control policies” In Learning for Dynamics and Control, 2020, pp. 361–373 PMLR
- “Differentiable convex optimization layers” In Advances in neural information processing systems 32, 2019
- Jérôme Bolte, Edouard Pauwels and Antonio José Silveti-Falls “Differentiating nonsmooth solutions to parametric monotone inclusion problems” In arXiv preprint arXiv:2212.07844, 2022
- “Nonsmooth implicit differentiation for machine-learning and optimization” In Advances in neural information processing systems 34, 2021, pp. 13537–13549
- Jacek Bochnak, Michel Coste and Marie-Françoise Roy “Real algebraic geometry” Springer Science & Business Media, 2013
- Amir Beck “First-order methods in optimization” SIAM, 2017
- “Convex Analysis and Monotone Operator Theory in Hilbert Spaces” Springer, 2017
- Michel Coste “An introduction to semialgebraic geometry” Istituti editoriali e poligrafici internazionali, 2000
- James B Rawlings “Tutorial: Model predictive control technology” In Proceedings of the 1999 American Control Conference (Cat. No. 99CH36251) 1, 1999, pp. 662–676 IEEE
- “Successive linearization based model predictive control of variable stiffness actuated robots” In 2017 IEEE international conference on advanced intelligent mechatronics (AIM), 2017, pp. 1774–1779 IEEE
- Ahmed Aboudonia, Annika Eichler and John Lygeros “Distributed model predictive control with asymmetric adaptive terminal sets for the regulation of large-scale systems” In IFAC-PapersOnLine 53.2 Elsevier, 2020, pp. 6899–6904
- Eric C Kerrigan and Jan M Maciejowski “Soft constraints and exact penalty functions in model predictive control” In UKACC International Conference (Control 2000), Cambridge, 2000
- James V Burke “An exact penalization viewpoint of constrained optimization” In SIAM Journal on control and optimization 29.4 SIAM, 1991, pp. 968–998
- “CasADi: a software framework for nonlinear optimization and optimal control” In Mathematical Programming Computation 11 Springer, 2019, pp. 1–36
- “qpOASES: A parametric active-set algorithm for quadratic programming” In Mathematical Programming Computation 6 Springer, 2014, pp. 327–363
- Gerd Wachsmuth “On LICQ and the uniqueness of Lagrange multipliers” In Operations Research Letters 41.1 Elsevier, 2013, pp. 78–80