Intelligent control through learning and optimization

AMATH / CSE 579

Winter 2014: SLN 10255, MW 3:30-4:50, EEB 026




Emo Todorov

Office: CSE 422
Email: todorov@cs.washington.edu


Course Description

Design of near-optimal controllers for complex dynamical systems, using analytical techniques, machine learning, and optimization. Topics from deterministic and stochastic optimal control, reinforcement learning and dynamic programming, numerical optimization in the context of control, and robotics. Prerequisite: vector calculus, linear algebra, and Matlab. Recommended: differential equations, stochastic processes, and optimization.


Homework

Homework 1

Optional Homework 2, due March 23

Code

MDP solver: all problem formulations and algorithms

Acrobot dynamics

Acrobot animation

Lecture slides

Introduction

Markov Decision Processes and Bellman Equations

Controlled Diffusions and Hamilton-Jacobi-Bellman Equations

Linearly-Solvable Stochastic Optimal Control Problems

Linearly-Quadratic-Gaussian Controllers and Kalman Filters

Research Talk

Pontryagin's Maximum Principle

Trajectory Optimization

Papers presented in class

Least-Squares Policy Iteration

An Analysis of Linear Models, Linear Value-Function Approximation, and Feature Selection for Reinforcement Learning

Value Function Approximation in Reinforcement Learning Using the Fourier Basis

Least Squares Solutions of the HJB Equation With Neural Network Value-Function Approximators

Reinforcement Learning In Continuous Time and Space

Proto-value Functions: A Laplacian Framework for Learning Representation and Control in Markov Decision Processes

Dynamical Movement Primitives: Learning Attractor Models forMotor Behaviors

An Iterative Path Integral Stochastic Optimal Control Approach for Learning Robotic Tasks

Policy Gradient Methods for Reinforcement Learning with Function Approximation

Compositionality of Optimal Control Laws

Policy Gradients in Linearly-Solvable MDPs

Robot Trajectory Optimization using Approximate Inference

A Reduction of Imitation Learning and Structured Prediction to No-Regret Online Learning

Natural Actor-Critic

SIMBICON: Simple Biped Locomotion Control

Optimizing Walking Controllers for Uncertain Inputs and Environments

Trajectory Optimization for Full-Body Movements with Complex Contacts

Dynamic Optimization of Human Walking

Computer Optimization of a Minimal Biped Model Discovers Walking and Running

Optimal Sensorimotor Transformations for Balance

Synthesis of Detailed Hand Manipulations Using Contact Smoothing

Continuous Character Control with Low-Dimensional Embeddings

Motion Fields for Interactive Character Animation

General Readings

A. Barto and R. Sutton (1998) Reinforcement learning: An introduction (online book)

E. Todorov (2006) Optimal control theory (book chapter)

D. Bertsekas (2008) Dynamic programming (lecture slides)

R. Tedrake (2009) Underactuated robotics: Learning, planning and control (lecture notes)

B. Van Roy (2004) Approximate dynamic programming (lecture notes)

P. Abbeel (2009) Advanced robotics (lecture slides)