On the Stability of Nonlinear Receding Horizon Control: A Geometric Perspective

@article{Westenbroek2021OnTS,
  title={On the Stability of Nonlinear Receding Horizon Control: A Geometric Perspective},
  author={Tyler Westenbroek and Max Simchowitz and M.I. Jordan and S. Shankar Sastry},
  journal={2021 60th IEEE Conference on Decision and Control (CDC)},
  year={2021},
  pages={742-749}
}
The widespread adoption of nonlinear Receding Horizon Control (RHC) strategies by industry has led to more than 30 years of intense research efforts to provide stability guarantees for these methods. However, current theoretical guarantees require that each (generally nonconvex) planning problem can be solved to (approximate) global optimality, which is an unrealistic requirement for the derivative-based local optimization methods generally used in practical implementations of RHC. This paper… 
Complexity Bounds of Iterative Linear Quadratic Optimization Algorithms for Discrete Time Nonlinear Control
TLDR
This work presents how the surjectivity of the linearized dynamics can be ensured by appropriate discretization schemes given the existence of a feedback linearization scheme, and uncovers several convergence phases for regularized generalized Gauss-Newton algorithms.
Online Control of Unknown Time-Varying Dynamical Systems
TLDR
It is proved a lower bound that no algorithm can obtain sublinear regret with respect to the class of Disturbance Response policies up to the aforementioned system variability term, and it is shown that offline planning over the state linear feedback policies is NP-hard, suggesting hardness of the online learning problem.
Provable Regret Bounds for Deep Online Learning and Control
TLDR
It is shown that over any sequence of convex loss functions, any low-regret algorithm can be adapted to optimize the parameters of a neural network such that it competes with the best net in hindsight.

References

SHOWING 1-10 OF 32 REFERENCES
Unconstrained receding-horizon control of nonlinear systems
TLDR
It is shown that the guaranteed region of operation contains that of the CLF controller and may be made as large as desired by increasing the optimization horizon (restricted, of course, to the infinite horizon domain).
Robust receding horizon control of constrained nonlinear systems
TLDR
This paper presents a method for the construction of a robust dual-mode, receding horizon controller which can be employed for a wide class of nonlinear systems with state and control constraints and model error, and requires considerably less online computation than existingReceding horizon controllers for nonlinear, constrained systems.
NONLINEAR OPTIMAL CONTROL: A CONTROL LYAPUNOV FUNCTION AND RECEDING HORIZON PERSPECTIVE
Two well known approaches to nonlinear control involve the use of control Lyapunov functions (CLFs) and receding horizon control (RHC), also known as model predictive control (MPC). The on‐line
Global Convergence of Policy Gradient Methods for the Linear Quadratic Regulator
TLDR
This work bridges the gap showing that (model free) policy gradient methods globally converge to the optimal solution and are efficient (polynomially so in relevant problem dependent quantities) with regards to their sample and computational complexities.
Constrained model predictive control: Stability and optimality
Receding horizon control of nonlinear systems
  • H. MichalskaD. Mayne
  • Mathematics
    Proceedings of the 28th IEEE Conference on Decision and Control,
  • 1989
TLDR
The authors show that receding horizon control stabilizes a wide class of nonlinear systems, extending the results of their previous paper, which require the finite-horizon value function to be continuously differentiable.
Suboptimal model predictive control (feasibility implies stability)
TLDR
Two suboptimal MPC schemes are presented and analyzed that are guaranteed to be stabilizing, provided an initial feasible solution is available and for which the computational requirements are more reasonable.
Adaptive control of linearizable systems
The authors give some initial results on the adaptive control of minimum-phase nonlinear systems which are exactly input-output linearizable by state feedback. Parameter adaptation is used as a
Output regulation of nonlinear systems
The problem of controlling a fixed nonlinear plant in order to have its output track (or reject) a family of reference (or disturbance) signal produced by some external generator is discussed. It is
How to Escape Saddle Points Efficiently
TLDR
This paper shows that a perturbed form of gradient descent converges to a second-order stationary point in a number iterations which depends only poly-logarithmically on dimension, which shows that perturbed gradient descent can escape saddle points almost for free.
...
...