Dynamical Systems and Control 76929 | Winter 2009-10 |
Schedule & Handouts: | ||||
---|---|---|---|---|
date | Subject | notes | Relevant material | |
19.10.09 | Introduction and math review, intro to linear systems | ppt pdf |
C-Leg_berlin_short.mpg C-Leg_slope_warped5.avi stand_up1.mov stand_up2.mov Bionic Arms Morimoto and Doya 2001 Wolpert-lab, TOPS project demo Harris & Wolpert 1998 Wolpert & Ghahramani 2000 |
|
26.10.09 | Phase plane analysis, singular points,
linearization near stationary points |
slides (ppt) |
Laplace Tutorial (1st and) 2nd order LTI system analysis |
|
2.11.09 | Simple PID control near stationary points. |
slides(pdf) notes (pdf) |
numeric integration movie |
|
22.11.09 | Autonomous LTI systems in matrix form: solution as matrix exponential, stability, discrete time counterpart |
slides (pdf) notes (pdf) |
laplace table Autonomous linear dynamical systems Solution via Laplace transform and matrix exponential Eigenvectors and diagonalization LTI systems & Laplace: chen 2.1-2.3 Linear Algebra: Chen, 3, Stengel, 2.1-2.2 |
|
22.11.09 | LTI systems with inputs: transfer functions, feedback | notes (pdf) | ||
Constrained optimization (Lagrange multipliers) |
slides (pdf) notes (pdf) |
|||
Block Diagram reduction, Signal Flow Graphs, Mason's rule Constrained optimization (Lagrange multipliers) |
Lecture slides (pdf, 4up) |
Block Diagram, SFG, Mason (10mb, pdf, acrobat 7) | ||
Controllability |
slides (pdf) notes (pdf) |
Stengel: Chapter 2.5 Chen: 4.1, 4.2, 4.2.1, 6.1-6.3 Controllability (Boyd) |
||
Least squares methods, SVD, pseudo inverse |
lecture slides notes( ps / pdf ) |
Least squares methods (Boyd) SVD (Boyd) |
||
SVD continued, Observability, State estimate, Gaussian noise |
lecture slides Notes(from last year. inc. Kalman Filter, not covered yet) ps / pdf ) |
|||
Kalman Filter & EKF |
slides (by Stephan Boyd):
estimation,
Kalman filter,
EKF |
Stengel: Ch 2.4
(pp. 86-114) Ch 4, 4.1, 4.2 (pp
299-322), Ch 4.3 (pp 342-347) Emery Brown's mouse tracking KF on Wikipedia |
||
Particle Filtering |
ps /
pdf slides(pdf) |
A tutorial on particle filtering (Arulampalam et al.) Introduction to Monte Carlo methods (Mackay) PF on Wikipedia | ||
Intro to variational methods | Bryson&Ho PP. 1-55 | |||
Variational mathods intro continued + Descrete time optimal control, LQR. | ps / pdf |
LQR via Lagrange multipliers (Boyd) LQR discrete time, finite horizon (Boyd) |
||
yet more variants of variational methods in optimal control. |
slides ps / pdf |
Bryson & Ho, Chap 2-3 (see slides for specific sections) | ||
HJB, Discrete version, Dynamic Programing, DP -> RL |
slides |
Bryson & Ho Chap 4. | ||
Inro to RL: DP, Monte Carlo |
ps /
pdf DP: Sutton & Barto, Chap. 4 Monte Carlo: Sutton & Barto, Chap. 5 ppt |
Sutton and Barto,Reinforcement learning, an Introduction (the book, at Sutton's web site) Kaelbling,Littman & Moore,Reinforcement Learning: A Survey,JAIR 1996 |
||
Intro to RL: TD(0) (Temporal Difference), Q-Learning | Sutton & Barto, Chap. 6, Sections 1-6,10,11 | W. Schultz,Predictive Reward Signal of Dopamin Neurons | ||
RL - TD(lambda) , RL in continuous space/time |
TD slides(ppt) Doya slides1(pdf) |
Sutton & Barto, Chap. 7, Sec. 1-5 K. Doya, Neural Computation 12,2000 |
||
Schedule & Handouts: | ||||
---|---|---|---|---|
date | teacher | Subject | notes | Relevant material |
Lavi | EM for linear dynamical systems |
ps /
pdf slides(pdf) |
A View of the EM Algorithm
That Justifies Incremental, Sparse and other Variants (Neal and Hinton)
Parameter estimation for LDS (Ghahramani and Hinton) |
|
7.1.07 | Lavi | Deriving system dynamics using Lagrange's Equations | slides |