Module C: Convex Optimization in Control

In a nutshell, control theory is the study of influencing trajectories of a dynamical

system to satisfy desired properties.

Static vs. Dynamic System:

Example:

State-space representation:

1
 Questions of Interest

Stability:

Identification:

State Estimation :

Optimal Control:

Robust Control:

2
 A. Discrete-time Optimal Control

Discrete-time State-Space Model:

Goal: Starting from an initial state z0 , compute a sequence of control inputs

(u0 , u1 , . . . , uT ) such that the state at time T , denoted zT = z des which is
the desired state.
Let us formulate an optimization problem to achieve this goal.
Decision Variables:

Cost Function:

Constraints:

3
 Finite-Horizon Optimal Control Problem

When is the above problem a convex optimization problem?

4
Discrete-time Linear Quadratic Regulation Problem

5
 B. Stability

Consider a continuous-time (autonomous) dynamical system: ẋ = f (x) with

initial state x0 .

Equilibrium point:

Stability of an equilibrium point:

Lyapunov Stability Theorem:

6
 Stability of a Continuous-time LTI System

An autonomous LTI System is stated as ẋ = Ax with initial state x0 .

7
 Linear Matrix Inequalities

Definition:

8
Primal and Dual forms of Optimization with LMI
Constraints

9
 Stability of a Discrete-time LTI System

An discrete-time autonomous LTI System is stated as xk+1 = Axk with

initial state x0 .

10