CS 4/5789: Introduction to Reinforcement Learning

Lecture 8: Linear Quadratic Regulator

Prof. Sarah Dean

MW 2:45-4pm
255 Olin Hall

Reminders

Homework this week
- Programming Assignment 1 due TONIGHT
- Next PSet and PA released tonight
  - PSet due next Wednesday
  - PA due in 2 weeks
My office hours:
- Tuesdays 10:30-11:30am in Gates 416A
- Wednesdays 4-4:50pm in Olin 255 (right after lecture)

Agenda

1. Recap

2. Linear Control

3. Linear Quadratic Regulator

Recap: Optimal Control

Continuous $\mathcal S = \mathbb R^{n_s}$ and $\mathcal A = \mathbb R^{n_a}$
Cost to be minimized $c=(c_0,\dots, c_{H-1}, c_H)$
Deterministic transitions described by dynamics function $$s_{t+1} = f(s_t, a_t)$$
Finite horizon $H$

$\mathcal M = \{\mathcal{S}, \mathcal{A}, c, f, H\}$

minimize $\displaystyle\sum_{t=0}^{H-1} c_t(s_t, a_t)+c_H(s_H)$

s.t. $s_{t+1}=f(s_t, a_t), ~~a_t=\pi_t(s_t)$

$\pi$

Recap: Linear Dynamics

The dynamics function $f$ has a linear form $$ s_{t+1} = As_t + Ba_t $$
$A$ describes the evolution of the state when there is no action (internal dynamics) $$ s_{t+1}=As_t$$

Recap: Trajectories and Stability

$0<\lambda_2<\lambda_1<1$

$0<\lambda_2<1<\lambda_1$

$1<\lambda_2<\lambda_1$

$\mathbb C$

$\mathcal R(\lambda)$

$\mathcal I(\lambda)$

Trajectory is determined by the eigenstructure of $A$

$s_1$

$s_2$

Recap: Trajectories and Stability

$\mathbb C$

$\mathcal R(\lambda)$

$\mathcal I(\lambda)$

Trajectory is determined by the eigenstructure of $A$

$s_1$

$s_2$

$\lambda = \alpha \pm i \beta$

$\mathbb C$

$\mathcal R(\lambda)$

$\mathcal I(\lambda)$

Trajectory is determined by the eigenstructure of $A$

$s_1$

$s_2$

$\lambda = \alpha \pm i \beta$

$0<\alpha^2+\beta^2<1$

$1<\alpha^2+\beta^2$

Recap: Trajectories and Stability

$\mathbb C$

$\mathcal R(\lambda)$

$\mathcal I(\lambda)$

Trajectory is determined by the eigenstructure of $A$

$s_1$

$s_2$

Recap: Trajectories and Stability

$\lambda_1 = \lambda_2=\lambda$

$\mathbb C$

$\mathcal R(\lambda)$

$\mathcal I(\lambda)$

Trajectory is determined by the eigenstructure of $A$

depends on if $A$ is diagonalizable

$s_1$

$s_2$

$0<\lambda<1$

$\lambda>1$

Recap: Trajectories and Stability

simulations

$\lambda_1 = \lambda_2=\lambda$

Recap: Stability Theorem

Theorem: Let $\{\lambda_i\}_{i=1}^n\subset \mathbb C$ be the eigenvalues of $A$.
Then for $s_{t+1}=As_t$, the equilibrium $s_{eq}=0$ is

asymptotically stable $\iff \max_{i\in[n]}|\lambda_i|<1$
unstable if $\max_{i\in[n]}|\lambda_i|> 1$
call $\max_{i\in[n]}|\lambda_i|=1$ "marginally (un)stable"

$\mathbb C$

Stability Theorem

Proof

If $A$ is diagonalizable, then any $s_0$ can be written as a linear combination of eigenvectors $s_0 = \sum_{i=1}^{n_s} \alpha_i v_i$
- By definition, $Av_i = \lambda_i v_i$
- Therefore, $s_t = \sum_{i=1}^{n_s}\alpha_i \lambda_i^t v_i$
- Thus $s_t\to 0$ if and only if all $|\lambda_i|<1$, and if any $|\lambda_i|>1$, $\|s_t\|\to\infty$
Proof in the non-diagonalizable case is out of scope, but it follows using the Jordan Normal Form

Marginally (un)stable

We call $\max_i|\lambda_i|=1$ "marginally (un)stable"
Consider independent investing example: (not unstable $\lambda_2<1$) $$ s_{t} = \begin{bmatrix} 1 &0 \\0 & \lambda_2 \end{bmatrix}^t s_0 $$
Consider UAV example: (unstable)$$s_{t} = \begin{bmatrix} 1 & 1 \\0 & 1 \end{bmatrix}^t s_0 =\begin{bmatrix} 1 & t\\ & 1\end{bmatrix} s_0 $$
Depends on eigenvectors not just eigenvalues!

Agenda

1. Recap

2. Linear Control

3. Linear Quadratic Regulator

Controlled Trajectories

Full dynamics depend on actions $$ s_{t+1} = As_t+Ba_t $$
The trajectories can be written as (PSet 3) $$ s_{t} = A^t s_0 + \sum_{k=0}^{t-1}A^k Ba_{t-k-1} $$
The internal dynamics $A$ determines the long term effects of actions

Example

Setting: hovering UAV over a target $$s_{t+1} = \begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix}s_t + \begin{bmatrix}0\\ 1\end{bmatrix}a_t$$
Initially at rest, then one rightward thrust followed by one leftward thrust $$a_0=1,\quad a_{t_0}=-1,\quad a_k=0~~k\notin\{0,t_0\} $$

$a_t$

$s_{t} = \displaystyle \begin{bmatrix}1 & t \\ 0 & 1\end{bmatrix}\begin{bmatrix}\mathsf{pos}_0 \\ 0 \end{bmatrix}+ \sum_{k=0}^{t-1} \begin{bmatrix}1 & k\\ 0 & 1\end{bmatrix} \begin{bmatrix}0\\ 1\end{bmatrix}a_{t-k-1}$
$s_{t} = \displaystyle \begin{bmatrix}\mathsf{pos}_0 \\ 0 \end{bmatrix}+ \begin{bmatrix}1 & t-1\\ 0 & 1\end{bmatrix} \begin{bmatrix}0\\ 1\end{bmatrix}- \begin{bmatrix}1 & t-t_0-1\\ 0 & 1\end{bmatrix} \begin{bmatrix}0\\ 1\end{bmatrix}$
for $t\leq t_0$, $s_{t} = \displaystyle \begin{bmatrix}\mathsf{pos}_0+ t-1 \\ 1 \end{bmatrix}$ and for $t\geq t_0$, $s_{t} = \displaystyle \begin{bmatrix}\mathsf{pos}_0+ t_0 \\ 0 \end{bmatrix}$

Linear Policy

Linear policy defined by $a_t=Ks_t$: $$ s_{t+1} = As_t+BKs_t = (A+BK)s_t$$
The trajectories can be written as $$ s_{t} = (A+BK)^t s_0 $$
The internal dynamics $A$ are modified depending on $B$ and $K$

Example

Setting: hovering UAV over a target $$s_{t+1} = \begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix}s_t + \begin{bmatrix}0\\ 1\end{bmatrix}a_t$$
Thrust according to distance from target $a_t = -(\mathsf{pos}_t- x)$

$a_t$

$s_{t+1} - \begin{bmatrix}x\\ 0\end{bmatrix} = \begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix}\left(s_t -\begin{bmatrix}x\\ 0\end{bmatrix}\right) + \begin{bmatrix}0\\ 1\end{bmatrix}a_t$
$\left(s_{t+1} - \begin{bmatrix}x\\ 0\end{bmatrix}\right) = \begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix}\left(s_t -\begin{bmatrix}x\\ 0\end{bmatrix}\right) + \begin{bmatrix}0\\ 1\end{bmatrix}\begin{bmatrix}-1& 0\end{bmatrix} \left(s_t -\begin{bmatrix}x\\ 0\end{bmatrix}\right)$
$\left(s_{t} - \begin{bmatrix}x\\ 0\end{bmatrix}\right) = \begin{bmatrix}1 & 1 \\ -1& 1\end{bmatrix}^t\left(s_0 -\begin{bmatrix}x\\ 0\end{bmatrix}\right)$

PollEV

Example

Setting: hovering UAV over a target $$s_{t+1} = \begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix}s_t + \begin{bmatrix}0\\ 1\end{bmatrix}a_t$$
Thrust according to distance from target $a_t = -(\mathsf{pos}_t+\mathsf{vel}_t- x)$

$a_t$

$\left(s_{t+1} - \begin{bmatrix}x\\ 0\end{bmatrix}\right) = \begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix}\left(s_t -\begin{bmatrix}x\\ 0\end{bmatrix}\right) + \begin{bmatrix}0\\ 1\end{bmatrix}\begin{bmatrix}-1& -1\end{bmatrix} \left(s_t -\begin{bmatrix}x\\ 0\end{bmatrix}\right)$
$\left(s_{t} - \begin{bmatrix}x\\ 0\end{bmatrix}\right) = \begin{bmatrix}1 & 1 \\ -1 & 0\end{bmatrix}^t\left(s_0 -\begin{bmatrix}x\\ 0\end{bmatrix}\right)$

simulations

Agenda

1. Recap

2. Linear Control

3. Linear Quadratic Regulator

Linear Quadratic Regulator

Special case of optimal control problem with

Quadratic cost $$c_t(s,a) = s^\top Qs+ a^\top Ra,\quad c_H = s^\top Qs$$
Linear dynamics $$s_{t+1} = As_t+ Ba_t$$

minimize $\displaystyle\sum_{t=0}^{H-1} s_t^\top Qs_t +a_t^\top Ra_t+s_H^\top Q s_H$

s.t. $s_{t+1}=As_t+B a_t, ~~a_t=\pi_t(s_t)$

$\pi$

Example

Setting: hovering UAV over a target
Action: thrust right/left
State is $s_t = \begin{bmatrix}\mathsf{position}_t - x\\ \mathsf{velocity}_t\end{bmatrix}$
$c_t(s_t, a_t) = (\mathsf{position}_t-x)^2+\lambda a_t^2$
$f(s_t, a_t) = \begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix}s_t + \begin{bmatrix}0\\ 1\end{bmatrix}a_t$

$a_t$

$Q = \begin{bmatrix}1&0\\ 0&0\end{bmatrix},\quad R=\lambda$

Example

Setting: hovering UAV over a target
Action: thrust right/left
State: distance from target, velocity
Consider $H=1$

$$\min_{a}\quad s^\top \begin{bmatrix}1&0\\ 0&0\end{bmatrix}s + (s')^\top \begin{bmatrix}1&0\\ 0&0\end{bmatrix}s' +\lambda a^2 \quad \text{s.t.} \quad s' = \begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix}s + \begin{bmatrix}0\\ 1\end{bmatrix}a $$

$$\min_{a}\quad (\begin{bmatrix}1&0\end{bmatrix}s)^2 + (\begin{bmatrix}1&1\end{bmatrix}s)^2 + \lambda a^2 \quad \implies a^\star = 0 $$

$a_t$

Example

$$\min_{a_0, a_1}\quad s_0^\top \begin{bmatrix}1&0\\ 0&0\end{bmatrix}s_0 + s_1^\top \begin{bmatrix}1&0\\ 0&0\end{bmatrix}s_1 + s_2^\top \begin{bmatrix}1&0\\ 0&0\end{bmatrix}s_2+\lambda a_{0}^2+\lambda a_1^2 $$

$$\text{s.t.} \quad s_{1} = \begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix}s_{0} + \begin{bmatrix}0\\ 1\end{bmatrix}a_{0} , \quad \quad s_{2} = \begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix}s_{1} + \begin{bmatrix}0\\ 1\end{bmatrix}a_{1} $$

$$\min_{a_0}\quad s_0^\top \begin{bmatrix}1&0\\ 0&0\end{bmatrix}s_0 + (\begin{bmatrix}1&0\end{bmatrix}s_1)^2 + (\begin{bmatrix}1&1\end{bmatrix}s_1)^2 +\lambda a_{0}^2 $$

$$\text{s.t.} \quad s_{1} = \begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix}s_{0} + \begin{bmatrix}0\\ 1\end{bmatrix}a_{0} , \quad $$

Setting: hovering UAV over a target
Action: thrust right/left
State: distance from target, velocity
Consider $H=2$

$a_t$

$$ a_0^\star = -\frac{\begin{bmatrix}1&2\end{bmatrix}s_0}{1+\lambda} $$

$a_1^\star=0$

$$\min_{a_0}\quad s_0^\top \begin{bmatrix}1&0\\ 0&0\end{bmatrix}s_0 + s_1^\top \begin{bmatrix}2&1\\ 1 & 1\end{bmatrix}s_1 +\lambda a_{0}^2 \quad \text{s.t.} \quad s_{1} = \begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix}s_{0} + \begin{bmatrix}0\\ 1\end{bmatrix}a_{0} , \quad $$

$$\min_{a_0}\quad s_0^\top \begin{bmatrix}1&0\\ 0&0\end{bmatrix}s_0 + \left(\begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix}s_{0} + \begin{bmatrix}0\\ 1\end{bmatrix}a_{0}\right)^\top \begin{bmatrix}2&1\\ 1 & 1\end{bmatrix}\left(\begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix}s_{0} + \begin{bmatrix}0\\ 1\end{bmatrix}a_{0}\right) +\lambda a_{0} ^2$$

$$\min_{a_0}\quad s_0^\top \begin{bmatrix}1&0\\ 0&0\end{bmatrix}s_0 + s_0^\top \begin{bmatrix}2&3\\ 3&5\end{bmatrix}s_0 + 2 s_0^\top \begin{bmatrix}1\\2\end{bmatrix}a_0 + a_0^2 +\lambda a_{0}^2 $$

$$\min_{a_0}\quad s_0^\top \begin{bmatrix}3&3\\ 3&5\end{bmatrix}s_0 + 2 s_0^\top \begin{bmatrix}1\\2\end{bmatrix}a_0 + (1 +\lambda)a_0^2 \implies a_0^\star = -\frac{\begin{bmatrix}1&2\end{bmatrix}s_0}{1+\lambda} $$

DP for Optimal Control

Reformulating for optimal control, our general purpose dynamic programming algorithm is:

Initialize $V^\star_H(s) = c_H(s)$
For $t=H-1, H-2, ..., 0$:
- $Q_t^\star(s,a) = c(s,a)+\mathbb E_{s'=f(s,a)}[V^\star_{t+1}(s')]$
- $\pi_t^\star(s) = \arg\min_a Q_t^\star(s,a)$
- $V^\star_{t}(s)=Q_t^\star(s,\pi_t^\star(s) )$
Return $\pi^\star = (\pi^\star_0,\dots ,\pi^\star_{H-1})$

$V^\star_{t+1}(f(s,a))$

LQR via DP

$V_H^\star(s) = s^\top Q s$
$t=H-1$: $\quad \min_{a} s^\top Q s+a^\top Ra+ (As+Ba)^\top Q (As+Ba)$
- $\quad \min_{a} s^\top (Q+A^\top QA) s+a^\top (R+B^\top QB) a+2s^\top A^\top Q Ba$
General minimization: $\arg\min_a c + a^\top M a + 2m^\top a$
- $2Ma_\star + 2m = 0 \implies a_\star = -M^{-1} m$
  - $ \pi_{H-1}^\star(s)=-(R+B^\top QB)^{-1}B^\top QAs$
- minimum is $c-m^\top M^{-1} m$
  - $V_{H-1}^\star(s) = s^\top (Q+A^\top QA - A^\top QB(R+B^\top QB)^{-1}B^\top QA) s$

DP: $V_t^\star (s) = \min_{a} c(s, a)+V_{t+1}^\star (f(s,a))$

LQR via DP

$V_H^\star(s) = s^\top Q s$
$t=H-1$: $\quad \min_{a} s^\top Q s+a^\top Ra+ (As+Ba)^\top Q (As+Ba)$
- $ \pi_{H-1}^\star(s)=-(R+B^\top QB)^{-1}B^\top QAs$
- $V_{H-1}^\star(s) = s^\top (Q+A^\top QA - A^\top QB(R+B^\top QB)^{-1}B^\top QA) s$

Theorem: For $t=0,\dots ,H-1$, the optimal value function is quadratic and the optimal policy is linear$$V^\star_t (s) = s^\top P_t s \quad\text{ and }\quad \pi_t^\star(s) = K_t s$$

where the matrices are defined as $P_{H} = Q$ and

$P_t = Q+A^\top P_{t+1}A - A^\top P_{t+1}B(R+B^\top P_{t+1}B)^{-1}B^\top P_{t+1}A$
$K_t = -(R+B^\top P_{t+1}B)^{-1}B^\top P_{t+1}A$

LQR Proof

Base case: $V_H^\star(s) = s^\top Q s$
Inductive step: Assume true at $t+1$.
DP at $t$: $V_t^\star(s)= \min_{a} s^\top Q s+a^\top Ra+ (As+Ba)^\top P_{t+1} (As+Ba)$
- $\quad \min_{a} s^\top (Q+A^\top P_{t+1}A) s+a^\top (R+B^\top P_{t+1} B) a+2s^\top A^\top P_{t+1} Ba$
General minimization: $\arg\min_a c + a^\top M a + 2m^\top a$ gives $a_\star = -M^{-1} m$ and minimum is $c-m^\top M^{-1} m$
- $ \pi_{t}^\star(s)=-(R+B^\top P_{t+1}B)^{-1}B^\top P_{t+1}As$
- $V_{t}^\star(s) = s^\top (Q+A^\top P_{t+1}A - A^\top P_{t+1}B(R+B^\top P_{t+1}B)^{-1}B^\top P_{t+1}A) s$

Example

Setting: hovering UAV over a target
Action: thrust right/left
State: distance from target, velocity
LQR$\left(\begin{bmatrix}1 & 1 \\ 0 & 1\end{bmatrix},\begin{bmatrix}0\\ 1\end{bmatrix},\begin{bmatrix}1&0\\ 0&0\end{bmatrix},\frac{1}{2}\right)$

$a_t$

$\pi_t^\star(s) = \begin{bmatrix}{ \gamma^\mathsf{pos}_t }& {\gamma_t^\mathsf{vel}} \end{bmatrix}s$

$\gamma^\mathsf{pos}$

$\gamma^\mathsf{vel}$

$-1$

$t$

$H$

LQR Extensions

The same dynamic programming method extends in a straightforward manner when:
1. Dynamics and costs are time varying
2. Affine term in the dynamics, cross terms in the costs
General form: $$c_t(s,a) = s^\top Q_t s+a^\top R_t a + s^\top M_ta + m_t$$ $$ f_t(s_t,a_t) = A_ts_t + B_t a_t +c_t$$
Many applications can be reformulated this way:
- e.g. trajectory tracking $c_t(s,a) = \|s-\bar s_t\|_2^2 + \|a\|_2^2$ for given $\bar s_t$
- Next lecture: general (nonlinear) dynamics and costs

Recap

PA 1 due TONIGHT

Linear Control
LQR

Next lecture: Nonlinear Control

Sp23 CS 4/5789: Lecture 8

By Sarah Dean

Sp23 CS 4/5789: Lecture 8

Sarah Dean PRO

asst prof in CS at Cornell

sdean.website

CS 4/5789: Introduction to Reinforcement Learning

Lecture 8: Linear Quadratic Regulator

Reminders

Agenda

Recap: Optimal Control

Recap: Linear Dynamics

Recap: Trajectories and Stability

Recap: Trajectories and Stability

Recap: Trajectories and Stability

Recap: Trajectories and Stability

Recap: Trajectories and Stability

Recap: Stability Theorem

Stability Theorem

Marginally (un)stable

Agenda

Controlled Trajectories

Example

Linear Policy

Example

Example

Agenda

Linear Quadratic Regulator

Example

Example

Example

DP for Optimal Control

LQR via DP

LQR via DP

LQR Proof

Example

LQR Extensions

Recap

Sp23 CS 4/5789: Lecture 8

More from Sarah Dean