This paper presents an alternative approach based on results from robust optimization to solve the stochastic linear-quadratic control (SLQC) problem, and considers a tight, second-order cone approximation to the SDP that can be solved much more efficiently when the problem has additional constraints.
Abstract:
Despite the celebrated success of dynamic programming for optimizing quadratic cost functions over linear systems, such an approach is limited by its inability to tractably deal with even simple constraints. In this paper, we present an alternative approach based on results from robust optimization to solve the stochastic linear-quadratic control (SLQC) problem. In the unconstrained case, the problem may be formulated as a semidefinite optimization problem (SDP). We show that we can reduce this SDP to optimization of a convex function over a scalar variable followed by matrix multiplication in the current state, thus yielding an approach that is amenable to closed-loop control and analogous to the Riccati equation in our framework. We also consider a tight, second-order cone (SOCP) approximation to the SDP that can be solved much more efficiently when the problem has additional constraints. Both the SDP and SOCP are tractable in the presence of control and state space constraints; moreover, compared to the Riccati approach, they provide much greater control over the stochastic behavior of the cost function when the noise in the system is distributed normally.
TL;DR: This paper surveys the primary research, both theoretical and applied, in the area of robust optimization (RO), focusing on the computational attractiveness of RO approaches, as well as the modeling power and broad applicability of the methodology.
TL;DR: In this article, the authors survey the primary research, both theoretical and applied, in the area of robust optimization and highlight applications of RO across a wide spectrum of domains, including finance, statistics, learning, and various areas of engineering.
TL;DR: In this paper, the authors survey the primary research, both theoretical and applied, in the area of robust optimization and highlight applications of RO across a wide spectrum of domains, including finance, statistics, learning, and various areas of engineering.
TL;DR: A relaxed robust counterpart for general conic optimization problems that preserves the computational tractability of the nominal problem and allows us to provide a guarantee on the probability that the robust solution is feasible when the uncertain coefficients obey independent and identically distributed normal distributions.
TL;DR: The main ideas underlying SMPC are presented and different classifications of the available methods are proposed in terms of the dynamic characteristics of the system under control, the performance index to be minimized, the meaning and management of the probabilistic constraints adopted, and their feasibility and convergence properties.
TL;DR: In this paper, the authors present a brief history of LMIs in control theory and discuss some of the standard problems involved in LMIs, such as linear matrix inequalities, linear differential inequalities, and matrix problems with analytic solutions.
TL;DR: The leading and most up-to-date textbook on the far-ranging algorithmic methododogy of Dynamic Programming, which can be used for optimal control, Markovian decision problems, planning and sequential decision making under uncertainty, and discrete/combinatorial optimization.
TL;DR: The authors present the basic theory of state-of-the-art polynomial time interior point methods for linear, conic quadratic, and semidefinite programming as well as their numerous applications in engineering.
Q1. What have the authors contributed in "Constrained stochastic lqc: a tractable approach" ?
In this paper, the authors present an alternative approach based on results from robust optimization to solve the stochastic linear-quadratic control ( SLQC ) problem. The authors show that they can reduce this SDP to optimization of a convex function over a scalar variable followed by matrix multiplication in the current state, thus yielding an approach that is amenable to closed-loop control and analogous to the Riccati equation in their framework. The authors also consider a tight, second-order cone ( SOCP ) approximation to the SDP that can be solved much more efficiently when the problem has additional constraints. Both the SDP and SOCP are tractable in the presence of control and state space constraints ; moreover, compared to the Riccati approach, they provide much greater control over the stochastic behavior of the cost function when the noise in the system is distributed normally.
Q2. What is the cost-to-go function in Proposition 10?
Proposition 10: With noisy estimates of the state given by (54), the cost-to-go can be written in the form(55)where , and are as in Proposition 2, is as in Proposition 2 with replacing , andFurthermore, the matrix is positive semidefinite.
Q3. What does the traditional approach assume is a random variable?
Most approaches assume is a random variable possessing some distributional properties and proceed to minimize (2) in an expected value sense.
Q4. What is the common way to ensure that the constraints on can be enforced?
B. Probabilistic State GuaranteesSince the state of the system is not exactly known, any constraints on can only be enforced in a probabilistic sense.
Q5. How can the authors solve the problem of a conic quadratic inequality?
In other words, the authors desire to find such thatBen-Tal and Nemirovski show that in the case of an ellipsoidal uncertainty set, the problem of optimizing over an uncertain conic quadratic inequality may be solved tractably using semidefinite programming.
Q6. What is the purpose of the probability results in Section III?
While it is true that this model does not seem, at first glance, to apply to random variables which are unbounded, the purpose of the probability results within this section is to show that the optimal solutions based their uncertainty model do in fact have reasonable performance guarantees even when the underlying disturbance vectors obey a different uncertainty model, namely, one admitting a probabilistic description.
Q7. What is the probability boundwhere for linear programs?
In the model of uncertainty in (10), when the authors use the -norm, i.e., , and under the assumption that , the authors have the probability boundwhere for linear programs (LPs), for SOCPs, and for SDPs ( is the dimension of the matrix in the SDP).
Q8. What is the performance of the SOCP?
Typically this runs longer than the SOCP, solidifying their assertion that the SOCP is much more suitable to efficient, closed-loop control.D. Performance on a Problem With Constraints