Professional Documents
Culture Documents
Necmiye Ozay
Outline:
• Course administration
• Course overview and goals
• Intro to hybrid system types
1
Course Administration
• Instructor: Necmiye Ozay
• Email: necmiye@umich.edu
• Office hours: MW 4:30-5:30pm,
or by appointment
• Course website: Canvas
• Grading:
– 40% HWs (7-8 assignments)
– 12% Paper presentation
– 8% Paper critiques (4 critiques)
– 5% Class participation
– 35% Project
Prerequisites: fully understanding the EECS
560 (AERO 550) (ME 564) material!
Knowledge of Lyapunov theory.
2
Course Administration
• Instructor: Necmiye Ozay • Term Project
• Email: necmiye@umich.edu - Ideally teams of three
• Office hours: MW 4:30-5:30pm, - Substantial part of the course
or by appointment - Some suggestions are included in
the syllabus
• Course website: Canvas
§ Applications
• Grading: § Software development
– 40% HWs (7-8 assignments) § Comparison of different
– 12% Paper presentation tools/techniques from the
– 8% Paper critiques (4 critiques) literature
– 5% Class participation § Investigation of an open problem
– 35% Project - In class poster presentation in
December, final report due finals week
• If you are auditing the course and want
to do a project, let me know.
A very good project can
possibly lead to a
conference/journal paper. 3
Course Administration
• Instructor: Necmiye Ozay • Homework
• Email: necmiye@umich.edu - Combination of problems and
computer exercises
• Office hours: MW 4:30-5:30pm,
or by appointment - one 48-hour grace period, one 48-
hour diminishing late penalty period
• Course website: Canvas
• Paper presentations and critiques:
• Grading:
- ~10 papers through the term
– 40% HWs (7-8 assignments)
- A team of two will be responsible for
– 12% Paper presentation
each paper:
– 8% Paper critiques (4 critiques)
§ Sketch the preliminaries, definitions
– 5% Class participation on the board
– 35% Project § Present the paper
§ Lead the discussion
§ 12% of your grade
- Rest: write a 1-2 page(s) critique of
the paper (due at the beginning of
the class), 8% of your grade 4
Couple of Announcements
• Problem Set #1 will be posted tonight*– due
9/10 (Thu)
• Two questionnaires:
– Course background survey (will be posted later
tonight*)
• Paper list will be posted on Canvas sometime
next week*
6
Schedule for the first few weeks
* Course overview
Week 1 Mon (8/28) * Examples of hybrid systems (hybrid automata)
Wed (8/30) * Lyapunov review
Observations:
• Typically large number of (discrete &
continuous) states and decision
variables
• Current design process is ad hoc
• Scalable tools for control design and
verification (theory and software) are
lagging
8
Current Practice
• Current control design process for
complex systems:
• Given some specs (plain English)
use engineering intuition,
experience and extensive
simulation/testing/fine-tuning to
come up with a solution
9
Current Practice
• Current control design process for
complex systems:
• Given some specs (plain English)
use engineering intuition,
experience and extensive
simulation/testing/fine-tuning to
come up with a solution
• little or no formal guarantees on
correctness: may find bugs/errors
via simulations but cannot prove
correctness
• no formal insight as to internal
mechanisms
10
Course Overview
• Model-based approach based
on rigorous control theory and Model-based approach
some computer science
• If you have taken any control
class, you already know how to
do this for simpler systems and
more “standard”
specifications!
11
Course Overview
• Model-based approach based
on rigorous control theory and Model-based approach
some computer science
• If you have taken any control
class, you already know how to
do this for simpler systems and
more “standard”
specifications!
• We will expand the types of
systems (e.g., hybrid) and the
types of specs (e.g., safety,
reachability, liveness) used for
analysis and synthesis.
• We will also look into some recent papers on what formal analysis can be made
when models or controllers are learned from data. 12
Course Overview
Course Goals:
• Provide a working knowledge of several analysis and synthesis
techniques/tools applicable to a larger class of systems (e.g., hybrid) with a
richer class of specs (e.g., safety, reachability, liveness)
• Provide a basic background in computer science and control theory that
can be used as a basis for further study
• Provide critical reading skills to follow the hybrid systems literature
14
Examples of Hybrid Systems and
Hybrid Models
Ex 1. Physics: Bouncing ball
17
Examples of Hybrid Systems and
Hybrid Models
Ex 4. Thermostat
19