Scribd
Upload
92 views10 pages

Optimal Control

This document introduces optimal control problems and methods. It discusses applications like rocket landing and dynamic walking. The optimal control problem aims to find control inputs to minimize a cost function while satisfying system dynamics and boundary conditions. Solutions include the linear-quadratic regulator (LQR) for linear systems. LQR provides a closed-form optimal control law using Riccati equations. Extensions cover infinite horizon, time-invariant systems and affine and tracking problems.

Uploaded by

Mayank Singal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
92 views10 pages

Optimal Control

This document introduces optimal control problems and methods. It discusses applications like rocket landing and dynamic walking. The optimal control problem aims to find control inputs to minimize a cost function while satisfying system dynamics and boundary conditions. Solutions include the linear-quadratic regulator (LQR) for linear systems. LQR provides a closed-form optimal control law using Riccati equations. Extensions cover infinite horizon, time-invariant systems and affine and tracking problems.

Uploaded by

Mayank Singal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Introduction to

Optimal Control
Yifan Hou
11/14/2018
• Rocket Landing 𝑓
• Thrust ~ fuel consumption
𝑚
𝑓
ℎ̈ = −𝑔 𝑚̇ = −𝑘𝑓 𝑔
𝑚
• Find control f to minimize fuel ℎ
consumption

• Dynamic walking
• Stabilize the inverted pendulum

• Find COM trajectory to move to goal


as fast as possible, while being
stable
The Optimal Control Problem
• System dynamics

• Cost function
Advantages:
Cost design is intuitive!

• Boundary conditions (optional)


Disadvantages:
(usually) no closed form solution

• Usually does not have closed form solution


• Special case: LQR
Linear-Quadratic Problem (Basic form)
• System dynamics

• Cost function

• Boundary conditions

• F and Q are not negative-definite,


• R is positive-definite
Solution: finite horizon, TV-LQR
• The necessary and sufficient condition for optimal control is:

Where is the solution to the following equation:


Riccati Equation

subjected to boundary condition .


Extension I: infinite horizon
• System dynamics

• Cost function

• Boundary conditions If the system is


controllable, then
• Q is not negative-definite, solution exists for the
optimal control problem
• R is positive-definite
Solution: infinite horizon TI-LQR
• The necessary and sufficient condition for optimal control is:

Where is the solution to the following equation:

Algebraic Riccati Equation


Stability of the closed-loop system
• Optimality stability !!

• Theorem

Decompose . If is observable,
then the closed loop system under optimal
control law is asymptotically stable.
Extension II: Affine system
• System dynamics
Extension III: Tracking Problem
• Goal trajectory: ∗ .
• Cost function: Denote ∗

You might also like