15-424/15-624: Foundations of Cyber-Physical Systems

01: Overview

André Platzer
[email protected]
Carnegie Mellon University, Pittsburgh, PA

http://symbolaris.com/course/fcps14.html
http://www.cs.cmu.edu/~aplatzer/course/fcps14.html

0.5
0.4
0.3
0.2
0.1
1.0
0.8
0.6
0.4
0.2

André Platzer (CMU) FCPS/01: Overview FCPS 1 / 24

 Outline
1 CPS: Introduction
Hybrid Systems & Cyber-Physical Systems
Applications
Robot Labs
2 15-424: Foundations of Cyber-Physical Systems
Objectives
Outline
Assessment
Labs
Resources
3 Approach
CPS Contracts
CPS Logic
Differential Dynamic Logic Family
4 Summary
André Platzer (CMU) FCPS/01: Overview FCPS 1 / 24
Can you trust a computer to control physics?

André Platzer (CMU) FCPS/01: Overview FCPS 2 / 24

 Can you trust a computer to control physics?

Rationale
1 Safety guarantees require analytic foundations
2 Foundations revolutionized digital computer science & society
3 Need even stronger foundations when software reaches out into our
physical world

Cyber-physical Systems
CPS combine cyber capabilities with physical capabilities to solve problems
that neither part could solve alone.

How can we provide people with cyber-physical systems they can bet their
lives on? – Jeannette Wing

André Platzer (CMU) FCPS/01: Overview FCPS 2 / 24

 Can you trust a computer to control physics?

Rationale
1 Safety guarantees require analytic foundations
2 Foundations revolutionized digital computer science & society
3 Need even stronger foundations when software reaches out into our
physical world

Cyber-physical Systems
CPS combine cyber capabilities with physical capabilities to solve problems
that neither part could solve alone.

1 CPS Foundations: intellectual grand challenge

2 Research & Industry applications abound
André Platzer (CMU) FCPS/01: Overview FCPS 2 / 24
 Outline
1 CPS: Introduction
Hybrid Systems & Cyber-Physical Systems
Applications
Robot Labs
2 15-424: Foundations of Cyber-Physical Systems
Objectives
Outline
Assessment
Labs
Resources
3 Approach
CPS Contracts
CPS Logic
Differential Dynamic Logic Family
4 Summary
André Platzer (CMU) FCPS/01: Overview FCPS 2 / 24
 CPS Analysis: Car Control

Challenge (Hybrid Systems)

Fixed rule describing state
evolution with both
Discrete dynamics
(control decisions)
Continuous dynamics
(differential equations)

v x
a 3.0 7
2
6
2.5
5
m
1 2.0
4
1.5
0
1 2 3 4 5
t 3
1.0
2
-1 0.5 1

5t 0
-2
0.0
0 1 2 3 4
0
1 2 3 4 5
t

André Platzer (CMU) FCPS/01: Overview FCPS 3 / 24

 Hybrid Systems & Cyber-Physical Systems

Mathematical model for complex physical systems:

Definition (Hybrid Systems)

systems with interacting discrete and continuous dynamics

Technical characteristics:
Definition (Cyber-Physical Systems)
(Distributed network of) computerized control for physical system
Computation, communication and control for physics

André Platzer (CMU) FCPS/01: Overview FCPS 4 / 24

 What CPS are around us?

What CPS will be around us in the future?

Which CPS do we trust with our lives?

André Platzer (CMU) FCPS/01: Overview FCPS 5 / 24

Successful CPS Proofs

(rx , ry )

fy
(vx , vy )

ey

(lx , ly ) xb ex fx

FM’11,LMCS’12,ICCPS’12,ITSC’11,ITSC’13,IJCAR’12
André Platzer (CMU) FCPS/01: Overview FCPS 6 / 24
 CPS Analysis & Design: Robot Lab

Challenge (Hybrid Systems)

Design & verify controller for 3.5

a robot avoiding obstacles 3.0

2.5

Accelerate / brake 2.0

(discrete dynamics) 1.5

1.0
1D motion
0.5

(continuous dynamics) 0.0

0 1 2 3 4 5 6

a v p
0.2
0.8 5
0.1
4
0.6
2 4 6 8 10 t 3
-0.1 0.4 px
2
-0.2 0.2
1 py

10 t
-0.3
2 4 6 8
2 4 6 8 10 t
André Platzer (CMU) FCPS/01: Overview FCPS 7 / 24
 CPS Analysis & Design: Robot Lab

Challenge (Hybrid Systems)

Design & verify controller for 3.5

a robot avoiding obstacles 3.0

2.5

Accelerate / brake 2.0

(discrete dynamics) 1.5

1.0
1D motion
0.5

(continuous dynamics) 0.0

0 1 2 3 4 5 6

a Ω d
0.2 1.0
0.00008 dx
0.1 0.8
0.00006
2 4 6 8 10 t 0.6
0.00004
-0.1 0.4
0.00002
-0.2 0.2

2 4 6 8 10 t
-0.3
2 4
dy
6 8 10 t
André Platzer (CMU) FCPS/01: Overview FCPS 7 / 24
 CPS Analysis & Design: Robot Lab

Challenge (Hybrid Systems)

Design & verify controller for 3.5

a robot avoiding obstacles 3.0

2.5

Accelerate / brake / stop 2.0

(discrete dynamics) 1.5

1.0
1D motion
0.5

(continuous dynamics) 0.0

0 1 2 3 4 5 6

a v p
2 4 6 8 10 t1.2 3.5 px
1.0 3.0
-0.05
0.8 2.5
2.0
-0.10 0.6
1.5
0.4
-0.15 1.0 py
0.2 0.5
-0.20
2 4 6 8 10 t 2 4 6 8 10 t
André Platzer (CMU) FCPS/01: Overview FCPS 8 / 24
 CPS Analysis & Design: Robot Lab

Challenge (Hybrid Systems)

Design & verify controller for 3.5

a robot avoiding obstacles 3.0

2.5

Accelerate / brake / stop 2.0

(discrete dynamics) 1.5

1.0
1D motion
0.5

(continuous dynamics) 0.0

0 1 2 3 4 5 6

a Ω d
2 4 6 8 10 t0.00012 1.0
0.00010
dx
-0.05 0.5
0.00008

-0.10 0.00006
2 4 6
dy
8 10 t
0.00004
-0.15 -0.5
0.00002

-0.20 2 4 6 8 10 t-1.0

André Platzer (CMU) FCPS/01: Overview FCPS 8 / 24

 CPS Analysis & Design: Robot Lab

Challenge (Hybrid Systems)

Design & verify controller for 3.5

a robot avoiding obstacles 3.0

2.5

Accelerate / brake 2.0

(discrete dynamics) 1.5

1.0
1D motion
0.5

(continuous dynamics) 0.0

0 1 2 3 4 5 6

a v p
0.2
2.5 px
0.8
0.1
2.0

2 4 6 8 10 t0.6 1.5
-0.1 0.4
1.0 py
-0.2 0.2 0.5

-0.3
2 4 6 8 10 t 2 4 6 8 10 t
André Platzer (CMU) FCPS/01: Overview FCPS 9 / 24
 CPS Analysis & Design: Robot Lab

Challenge (Hybrid Systems)

Design & verify controller for 3.5

a robot avoiding obstacles 3.0

2.5

Accelerate / brake 2.0

(discrete dynamics) 1.5

1.0
1D motion
0.5

(continuous dynamics) 0.0

0 1 2 3 4 5 6

a Ω d
0.2 1.0
0.00008 dx
0.1 0.8
0.00006
2 4 6 8 10 t 0.6
0.00004
-0.1 0.4
0.00002
-0.2 0.2

2 4 6 8 10 t
-0.3
2 4
dy
6 8 10 t
André Platzer (CMU) FCPS/01: Overview FCPS 9 / 24
 CPS Analysis & Design: Robot Lab

Challenge (Hybrid Systems)

Design & verify controller for 3.5

a robot avoiding obstacles 3.0

2.5

Accel / brake / steer 2.0

(discrete dynamics) 1.5

1.0
2D motion
0.5

(continuous dynamics) 0.0

0 1 2 3 4 5 6

a v p
1.0
0.2 8
0.8
2 4 6 8 10 t 6
0.6
-0.2 px
0.4 4
-0.4

-0.6 0.2 2
py
10 t
-0.8
2 4 6 8
2 4 6 8 10 t
André Platzer (CMU) FCPS/01: Overview FCPS 10 / 24
 CPS Analysis & Design: Robot Lab

Challenge (Hybrid Systems)

Design & verify controller for 3.5

a robot avoiding obstacles 3.0

2.5

Accel / brake / steer 2.0

(discrete dynamics) 1.5

1.0
2D motion
0.5

(continuous dynamics) 0.0

0 1 2 3 4 5 6

a Ω d
1.0
0.2 dx
0.5

2 4 6 8 10 t 0.5
-0.2 2 4 6 8 10 t
-0.4
-0.5
2 4 dy 6 8 10 t
-0.6 -1.0
-0.5
-0.8

André Platzer (CMU) FCPS/01: Overview FCPS 10 / 24

 CPS Analysis & Design: Robot Lab

Challenge (Hybrid Systems)

Design & verify controller for 3.5

a robot avoiding obstacles 3.0

2.5

Dynamic obstacles 2.0

(other agents) 1.5

1.0
Avoid collisions
0.5

(define safety) 0.0

0 1 2 3 4 5 6

a v p
2 4 6 8 10 t1.0 4
px
0.8
-1
3
0.6
-2
2
0.4

-3
0.2 1
py

10 t t
-4
2 4 6 8
2 4 6 8 10

André Platzer (CMU) FCPS/01: Overview FCPS 11 / 24

 CPS Analysis & Design: Robot Lab

Challenge (Hybrid Systems)

Design & verify controller for 3.5

a robot avoiding obstacles 3.0

2.5

Dynamic obstacles 2.0

(other agents) 1.5

1.0
Avoid collisions
0.5

(define safety) 0.0

0 1 2 3 4 5 6

a Ω d
2 4 6 8 10 t 1.0
dx
0.5
-1
0.5

2 4 6 8 10 t
-2

-0.5 2 4 6 8 10 t
-3 dy
-1.0 -0.5
-4

André Platzer (CMU) FCPS/01: Overview FCPS 11 / 24

 CPS Analysis & Design: Robot Lab

Challenge (Hybrid Systems)

Design & verify controller for 3.5

a robot avoiding obstacles 3.0

2.5

Control robot 2.0

(respect delays) 1.5

1.0
Environment interaction
0.5

(obstacles, agents, 0.0

0 1 2 3 4 5 6

uncertainty)
a v p
1.2 7
0.4
6
1.0
0.2
5
0.8
2 4 6 8 10 t 4
0.6 px
-0.2 3
0.4
-0.4 2
0.2
-0.6 1
py
2 4 6 8 10 t 2 4 6 8 10 t
André Platzer (CMU) FCPS/01: Overview FCPS 12 / 24
 CPS Analysis & Design: Robot Lab

Challenge (Hybrid Systems)

Design & verify controller for 3.5

a robot avoiding obstacles 3.0

2.5

Control robot 2.0

(respect delays) 1.5

1.0
Environment interaction
0.5

(obstacles, agents, 0.0

0 1 2 3 4 5 6

uncertainty)
a Ω d
1.0
0.4 dx
0.5
0.2
0.5
t 2 4 6 8 10 t dy
2 4 6 8 10
-0.2 -0.5
2 4 6 8 10 t
-0.4
-1.0
-0.6 -0.5

André Platzer (CMU) FCPS/01: Overview FCPS 12 / 24

 Proof Algebraic
Modal Computer
Theory Geometry
Logic Algebra
Theorem Differential
Proving Algebra

Model Lie
Checking Algebra
Logic Algebra

Closure Differential
Ordinals Equations

Fixpoints Logical Dynamical

& Lattices Systems
Foundations
Algorithms of Analysis
Cyber-Physical
Proof Differ-
Search Systems entiation

Decision Limit
Procedures Processes

Weierstraß Numerics Stochastics Stochastic

Approx- Differential
imation Equations

Hermite Differential
Interpolation Generators
Doob’s
Error Dynkin’s
Numerical Super-
Analysis Infinitesimal
Quadrature martingales
Generators

André Platzer (CMU) FCPS/01: Overview FCPS 13 / 24

 Outline
1 CPS: Introduction
Hybrid Systems & Cyber-Physical Systems
Applications
Robot Labs
2 15-424: Foundations of Cyber-Physical Systems
Objectives
Outline
Assessment
Labs
Resources
3 Approach
CPS Contracts
CPS Logic
Differential Dynamic Logic Family
4 Summary
André Platzer (CMU) FCPS/01: Overview FCPS 13 / 24
About 15-424/14-624 Foundations of CPS
Foundations!
Modeling & Control
1 Understand the core principles behind CPSs.
2 Develop models and controls.
3 Identify the relevant dynamical aspects.
Computational Thinking
1 Identify safety specifications and critical properties of CPSs.
2 Understand abstraction and system architectures.
3 Learn how to design by invariant.
4 Reason rigorously about CPS models.
5 Verify CPS models of appropriate scale.
CPS Skills
1 Understand the semantics of a CPS model.
2 Develop an intuition for operational effects.
3 Use higher-level model-predictive control.
Byproducts
1 Exposure to numerous math areas in action.
2 ...
André Platzer (CMU) FCPS/01: Overview FCPS 14 / 24
 Course Outline

1 Cyber-physical systems: introduction

2 Differential equations & domains
3 Choice & control
4 Safety & contracts
5 Dynamical systems & dynamic axioms
6 Truth & proof
7 Control loops & invariants
8 Events & delays
9 Differential equations & differential invariants
10 Differential equations & proofs
11 Dynamic logic & dynamical systems
12 Dynamical systems: discrete & continuous & hybrid

13 Robots / railway / air traffic / car CPS & applications

14 Virtual substitution & real arithmetic
15 Hybrid systems & games
André Platzer (CMU) FCPS/01: Overview FCPS 15 / 24
Assessment

Read Collaboration and Academic Integrity Policy Policy

≈22% Theory homework Due at beginning of lecture

≈51% Labs, including ≈22% final project Due at 22:00
Whitepaper For final project
Proposal For final project
Term paper Due with final project
≈11% Midterm
≈11% Final
≈5% Participation in class and in online comments

André Platzer (CMU) FCPS/01: Overview FCPS 16 / 24

 Robot Labs

1 Robot on Rails
a Autobots, Roll Out
b Charging Station
2 Robot on Highways
a with event-driven control
b with time-triggered control
3 Robot on Racetracks
a stay on the circular racetrack
b slow down to avoid collisions
4 Robot in a Plane
a with obstacle avoidance
b Robot vs. Roguebot: avoid collisions with moving obstacles
5 Robot in Star-lab: self-defined final project
6 Final project presented at CPS V&V Grand Prix CPS v&V Grand Prix

André Platzer (CMU) FCPS/01: Overview FCPS 17 / 24

 Resources

Prerequisites
15-122 Principles of Imperative Computation
21-122 Integration, Differential Equations, and Approximation
(21-241 Matrix algebra or
18-202 Mathematical Foundations of Electrical Engineering or
15-251 Great Theoretical Ideas in Computer Science)

You will be expected to follow extra background reading material as

needed.
Further reading and background material on the course web page
Check course web page periodically
http://symbolaris.com/course/fcps14.html
KeYmaera
Piazza
Autolab
Ask!
André Platzer (CMU) FCPS/01: Overview FCPS 18 / 24
Textbook

André Platzer.
Logical Analysis of Hybrid Systems.
Springer, 426p., 2010.
DOI 10.1007/978-3-642-14509-4
http://symbolaris.com/lahs/
CMU library e-book
André Platzer.
Foundations of Cyber-Physical Systems.
Lecture notes.
Computer Science Department
Carnegie Mellon University.
http://symbolaris.com/course/
fcps14-schedule.html

André Platzer (CMU) FCPS/01: Overview FCPS 19 / 24

 Outline
1 CPS: Introduction
Hybrid Systems & Cyber-Physical Systems
Applications
Robot Labs
2 15-424: Foundations of Cyber-Physical Systems
Objectives
Outline
Assessment
Labs
Resources
3 Approach
CPS Contracts
CPS Logic
Differential Dynamic Logic Family
4 Summary
André Platzer (CMU) FCPS/01: Overview FCPS 19 / 24
 CPS Design & CPS Contracts in Programs

HP Reveal in layers Contracts Reason about CPS

@ r e q u i r e s (0<=x & x=H & v =0)

@ r e q u i r e s ( g>0 & 1>=c>=0)
@ensures(0<=x & x<=H)
{
{x ’=v , v’=−g , x >=0};
i f ( x = 0) {
v := −c ∗ v ;
}
}∗ @ i n v a r i a n t ( 2 ∗ g ∗ x <= 2∗ g ∗H − v ˆ2 & x>=0)

CPS Simulate for intuition CT Design-by-invariant

André Platzer (CMU) FCPS/01: Overview FCPS 20 / 24

 CPS Contracts & CPS Logic

dL Logic for CPS Contracts Reason in Logic

0<=x & x=H & v=0

& g>0 & 1>=c>=0
−>
[{
{x ’=v , v’=−g , x >=0};
i f ( x = 0) {
v := −c ∗ v ;
}
}∗ @ i n v a r i a n t ( 2 ∗ g ∗ x <= 2∗ g ∗H − v ˆ2 & x>=0)
] (0<=x & x<=H)

CPS Analyze for precision CT Proof-by-invariant

André Platzer (CMU) FCPS/01: Overview FCPS 21 / 24

Family of Differential Dynamic Logics

e te c o n ti n

d i s cr

uous
al

stoc
i
rsar

nondet
has
e
adv

tic

JAR’08,CADE’11,LMCS’12,LICS’12,LICS’12
André Platzer (CMU) FCPS/01: Overview FCPS 22 / 24
 Family of Differential Dynamic Logics

differential dynamic logic

[α]φ φ
dL = DL + HP α

e te c o n ti n

d i s cr

uous
differential game logic stochastic differential DL
dGL = GL + HG SdL = DL + SHP
al

stoc
i
rsar

nondet
has
e
adv

tic
hαiφ hαiφ
φ φ

quantified differential DL
QdL = FOL + DL + QHP
JAR’08,CADE’11,LMCS’12,LICS’12,LICS’12
André Platzer (CMU) FCPS/01: Overview FCPS 22 / 24
 Proof Algebraic
Modal Computer
Theory Geometry
Logic Algebra
Theorem Differential
Proving Algebra

Model Lie
Checking Algebra
Logic Algebra

Closure Differential
Ordinals Equations

Fixpoints Logical Dynamical

& Lattices Systems
Foundations
Algorithms of Analysis
Cyber-Physical
Proof Differ-
Search Systems entiation

Decision Limit
Procedures Processes

Weierstraß Numerics Stochastics Stochastic

Approx- Differential
imation Equations

Hermite Differential
Interpolation Generators
Doob’s
Error Dynkin’s
Numerical Super-
Analysis Infinitesimal
Quadrature martingales
Generators

André Platzer (CMU) FCPS/01: Overview FCPS 23 / 24

 Outline
1 CPS: Introduction
Hybrid Systems & Cyber-Physical Systems
Applications
Robot Labs
2 15-424: Foundations of Cyber-Physical Systems
Objectives
Outline
Assessment
Labs
Resources
3 Approach
CPS Contracts
CPS Logic
Differential Dynamic Logic Family
4 Summary
André Platzer (CMU) FCPS/01: Overview FCPS 23 / 24
Can you trust a computer to control physics?

André Platzer (CMU) FCPS/01: Overview FCPS 24 / 24

 Can you trust a computer to control physics?

Rationale
1 Safety guarantees require analytic foundations
2 Foundations revolutionized digital computer science & society
3 Need even stronger foundations when software reaches out into our
physical world

Cyber-physical Systems
CPS combine cyber capabilities with physical capabilities to solve problems
that neither part could solve alone.

How can we provide people with cyber-physical systems they can bet their
lives on? – Jeannette Wing

André Platzer (CMU) FCPS/01: Overview FCPS 24 / 24

 Can you trust a computer to control physics?

Rationale
1 Safety guarantees require analytic foundations
2 Foundations revolutionized digital computer science & society
3 Need even stronger foundations when software reaches out into our
physical world

Cyber-physical Systems
CPS combine cyber capabilities with physical capabilities to solve problems
that neither part could solve alone.

1 CPS Foundations: intellectual grand challenge

2 Research & Industry applications abound
André Platzer (CMU) FCPS/01: Overview FCPS 24 / 24